Deployment on Producthunt daily

fizzy

Sat, 06 Dec 2025 15:27:23 +0800

basecamp/fizzy

Fizzy

This is the source code of Fizzy, the Kanban tracking tool for issues and ideas by 37signals.

Deploying Fizzy

If you’d like to run Fizzy on your own server, we recommend deploying it with Kamal. Kamal makes it easier to set up a bare server, copy the application to it, and manage the configuration settings that it uses.

(Kamal is also what we use to deploy Fizzy at 37signals. If you’re curious about what our deployment configuration looks like, you can find it inside fizzy-saas.)

This repo contains a starter deployment file that you can modify for your own specific use. That file lives at config/deploy.yml, which is the default place where Kamal will look for it.

The steps to configure your very own Fizzy are:

Fork the repo
Edit few things in config/deploy.yml, .kamal/secrets, and config/environments/production.rb
Run kamal setup to do your first deploy.

We’ll go through each of these in turn.

Fork the repo

To make it easy to customise Fizzy’s settings for your own instance, you should start by creating your own GitHub fork of the repo. That allows you to commit your changes, and track them over time. You can always re-sync your fork to pick up new changes from the main repo over time.

Once you’ve got your fork ready, run bin/setup from within it, to make sure everything is installed.

Editing the configuration

The config/deploy.yml has been mostly set up for you, but you’ll need to fill out some sections that are specific to your instance. To get started, the parts you need to change are all in the “About your deployment” section. We’ve added comments to that file to highlight what each setting needs to be, but the main ones are:

servers/web: Enter the hostname of the server you’re deploying to here. This should be an address that you can access via ssh.
ssh/user: If you access your server a root you can leave this alone; if you use a different user, set it here.
proxy/ssl and proxy/host: Kamal can set up SSL certificates for you automatically. To enable that, set the hostname again as host. If you don’t want SSL for some reason, you can set ssl: false to turn it off.
env/clear/MAILER_FROM_ADDRESS: This is the email address that Fizzy will send emails from. It should usually be an address from the same domain where you’re running Fizzy.

Fizzy also requires a few environment variables to be set up, some of which contain secrets. The simplest way to do this is to put them in a file called .kamal/secrets. Because this file will contain secret credentials, it’s important that you DON’T CHECK THIS FILE INTO YOUR REPO! You can add the filename to .gitignore to ensure you don’t commit this file accidentally.

If you use a password manager like 1Password, you can also opt to keep your secrets there instead. Refer to the Kamal documentation for more information about how to do that.

To store your secrets, create the file .kamal/secrets and enter something like the following:

SECRET_KEY_BASE=12345
VAPID_PUBLIC_KEY=something
VAPID_PRIVATE_KEY=somethingelse
SMTP_USERNAME=email-provider-username
SMTP_PASSWORD=email-provider-password

The values you enter here will be specific to you, and you can get or create them as follows:

SECRET_KEY_BASE should be a long, random secret. You can run bin/rails secret to create a suitable value for this.
VAPID_PUBLIC_KEY & VAPID_PRIVATE_KEY are a pair of credentials that are used for sending notifications. You can create your own keys by starting a development console with:
```
bin/rails c
```
And then run the following to create a new pair of keys:
1 2 3 4

vapid_key = WebPush.generate_key puts "VAPID_PRIVATE_KEY=#{vapid_key.private_key}" puts "VAPID_PUBLIC_KEY=#{vapid_key.public_key}"
SMTP_USERNAME/SMTP_PASSWORD are credentials you should get from your email provider.

Lastly, you’ll need to set up the rest of your email configuration in config/environments/production.rb. There is an example configuration in comments at the top of that file. The actual settings you use here will depend on your email provider, but in most cases will look similar to that section, so you can uncomment it and edit to suit. Note that it will use the SMTP_USERNAME and SMTP_PASSWORD values you entered in your secrets.

Once you’ve made all those changes, commit them to your fork so they’re saved.

Deploy Fizzy!

You can now do your first deploy by running:

bin/kamal setup

This will set up Docker (if needed), build your Fizzy app container, configure it, and start it running.

After the first deploy is done, any subsequent steps won’t need to do that initial setup. So for future deploys you can just run:

bin/kamal deploy

Development

Setting up

First, get everything installed and configured with:

1
2

bin/setup
bin/setup --reset # Reset the database and seed it

And then run the development server:

bin/dev

You’ll be able to access the app in development at http://fizzy.localhost:3006.

To login, enter david@example.com and grab the verification code from the browser console to sign in.

Web Push Notifications

Fizzy uses VAPID (Voluntary Application Server Identification) keys to send browser push notifications. For notifications to work in development you’ll need to generate a key pair and set these environment variables:

VAPID_PRIVATE_KEY
VAPID_PUBLIC_KEY

Generate them with the web-push gem:

vapid_key = WebPush.generate_key

puts "VAPID_PRIVATE_KEY=#{vapid_key.private_key}"
puts "VAPID_PUBLIC_KEY=#{vapid_key.public_key}"

Running tests

For fast feedback loops, unit tests can be run with:

bin/rails test

The full continuous integration tests can be run with:

bin/ci

Database configuration

Fizzy works with SQLite by default and supports MySQL too. You can switch adapters with the DATABASE_ADAPTER environment variable. For example, to develop locally against MySQL:

1
2

DATABASE_ADAPTER=mysql bin/setup --reset
DATABASE_ADAPTER=mysql bin/ci

The remote CI pipeline will run tests against both SQLite and MySQL.

Outbound Emails

You can view email previews at http://fizzy.localhost:3006/rails/mailers.

You can enable or disable letter_opener to open sent emails automatically with:

bin/rails dev:email

Under the hood, this will create or remove tmp/email-dev.txt.

SaaS gem

37signals bundles Fizzy with fizzy-saas, a companion gem that links Fizzy with our billing system and contains our production setup.

This gem depends on some private git repositories and it is not meant to be used by third parties. But we hope it can serve as inspiration for anyone wanting to run fizzy on their own infrastructure.

Contributing

We welcome contributions! Please read our style guide before submitting code.

License

Fizzy is released under the O’Saasy License.

Product Hunt Daily | 2025-10-13

Mon, 13 Oct 2025 13:09:03 +0000

1. Meku

Tagline: AI Web App and Site Builder
Description: Meku.dev is an AI-powered web app and site builder engineered for developers. Generate, customize and deploy full-stack web apps and sites from simple AI prompts. Comes with essential integrations and deployment tools and hosting to launch your MVP fast.
Website: open
Product Hunt: View on Product Hunt

Keyword: AI Web App Builder, Site Builder, AI, Web App, Full-Stack, Developer Tools, Deployment, Hosting, MVP, Meku.dev, AI Prompts
VotesCount: 🔺384
Featured: Yes
CreatedAt: 2025-10-12 07:01 AM (UTC)

2. Open SaaS 2.0

Tagline: Free, open-source SaaS starter kit with superpowers
Description: Build and launch your SaaS application faster with this starter kit—free, open-source, full-stack React + Node.js. Features include auth, payments, AI example app, and admin dashboard.
Website: open
Product Hunt: View on Product Hunt

Keyword: SaaS, open-source, React, Node.js, starter kit, free, auth, payments, AI, admin dashboard
VotesCount: 🔺335
Featured: Yes
CreatedAt: 2025-10-12 07:01 AM (UTC)

3. UNDOOMED

Tagline: Stop doomscrolling. A mute button for Reels, Shorts & feeds.
Description: Tired of “5 min” turning into 2 hours? Undoomed hides Reels/Shorts/TikTok feeds but leaves what matters—messages and posts. Per-app settings, quick switch, light “clarity” stats. iOS now, Android on the way.
Website: open
Product Hunt: View on Product Hunt

Keyword: doomscrolling, mute button, Reels, Shorts, feeds, iOS, Android, productivity, time management, social media, focus
VotesCount: 🔺258
Featured: Yes
CreatedAt: 2025-10-12 07:01 AM (UTC)

4. Getillustrations Figma Plugin

Tagline: Bring stunning vector illustrations straight into Figma
Description: The power of Getillustrations directly in Figma! Our plugin is packed with thousands of ready-to-use illustrations for UI, apps and websites. Fast, editable, and designed to keep your workflow smooth, just download and link to your All access account.
Website: open
Product Hunt: View on Product Hunt

Keyword: Figma plugin, vector illustrations, UI illustrations, app illustrations, website illustrations, editable illustrations, ready-to-use illustrations, Getillustrations
VotesCount: 🔺193
Featured: Yes
CreatedAt: 2025-10-12 07:01 AM (UTC)

5. AI Ultra Editing by Thumblify

Tagline: Edit Like a Pro
Description: ThumblifyAI Ultra Editing gives you full creative power. Lets you replace backgrounds, add elements, and fix every small detail in seconds. No limits, no complex tools. Just pure creativity made faster with AI.
Website: open
Product Hunt: View on Product Hunt

Keyword: AI Editing, Image Editing, Background Removal, AI, Creative Editing, Thumblify, Ultra Editing, Easy Editing, Fast Editing, AI Tools
VotesCount: 🔺173
Featured: Yes
CreatedAt: 2025-10-12 07:01 AM (UTC)

6. Monitoro

Tagline: Track AI & SEO Traffic o_O
Description: Track traffic from AI, Google, and Bing in one place. Finally!
Website: open
Product Hunt: View on Product Hunt

Keyword: AI traffic, SEO traffic, Google traffic, Bing traffic, traffic tracking, monitor
VotesCount: 🔺62
Featured: No
CreatedAt: 2025-10-12 07:01 AM (UTC)

7. Flow Tabs Pro

Tagline: Cut Chrome’s memory usage in half with smart tab suspension
Description: Stop Chrome from eating your RAM. Flow Tabs Pro suspends inactive tabs, cutting memory usage up to 50%. Super Focus Mode, session switching, and seamless tab management—all without disrupting your workflow. 7-day free trial.
Website: open
Product Hunt: View on Product Hunt

Keyword: Chrome, tab management, memory saver, tab suspension, RAM optimization, Flow Tabs Pro, productivity, free trial, Super Focus Mode, session switching
VotesCount: 🔺40
Featured: No
CreatedAt: 2025-10-12 07:01 AM (UTC)

8. Minarah - Prayer Reminder App

Tagline: Never miss a prayer on your mac
Description: Minarah is a full-screen prayer reminder for macOS with elegant alerts, Smart Countdown, and Focus Mode — helping you stay on time for every Salah without breaking focus.
Website: open
Product Hunt: View on Product Hunt

Keyword: Prayer reminder, macOS, Salah, alerts, countdown, focus mode, Islamic app, prayer times
VotesCount: 🔺25
Featured: No
CreatedAt: 2025-10-12 07:01 AM (UTC)

9. AI Financial Report (10K, Q, etc)

Tagline: Fast, clear annual report with smart financial metrics
Description: Delivers instant, comprehensive financial report summaries with one click. Upload statements and get a clear breakdown of key metrics business, finance statement, growth all displayed in a modern UI. Save hours, work smarter, and make informed decisions.
Website: open
Product Hunt: View on Product Hunt

Keyword: AI Financial Report, financial summaries, annual reports, 10K, Q reports, financial metrics, business analysis, fast reporting, modern UI, informed decisions, data analysis
VotesCount: 🔺19
Featured: No
CreatedAt: 2025-10-12 07:01 AM (UTC)

10. Strayl

Tagline: Build everywhere
Description: Strayl is a mobile AI coding agent - like Claude Code, but built for iOS. Edit, refactor, and preview your projects with AI. It connects to GitHub, indexes your codebase, connects to MCP servers to use their tools, and can spin up instant preview deployments.
Website: open
Product Hunt: View on Product Hunt

Keyword: AI coding agent, iOS, Code editing, Refactoring, AI-powered, GitHub integration, Instant preview, Deployment, Mobile coding, Strayl
VotesCount: 🔺19
Featured: No
CreatedAt: 2025-10-12 07:01 AM (UTC)

examples

Thu, 11 Sep 2025 15:29:44 +0800

vercel/examples

Vercel Examples

Solutions – Demos, reference architecture, and best practices
Starter – Functional applications which can act as a starting point
And more!

Vercel Templates

Multiple examples are being featured in Vercel’s Templates, visit that page for more advanced filtering options.

For Vercelians

Examples that have front matter metadata will create a new Draft template in Contentful, for more steps on how to publish a template, read Publishing Templates.

Adding a new example

To quickly start contributing with a new example, run the following commands:

1
2

pnpm i
pnpm new-example

If the script above isn’t used, make sure the example complies with the following:

It must have a .gitignore similar to plop-templates/example/.gitignore
It must have a package.json similar to plop-templates/example/package.json (usage of Next.js is optional). The license should be MIT
It must have a README.md similar to plop-templates/example/README.md. The example has to be able to include a demo URL (the Vercel team will deploy it!) and if it requires environment variables, it must have a .env.example file and instructions on how to set them up. Take bot-protection-datadome as an example.
- To customize the Vercel Deploy Button take a look at the docs, useful if the deployment has required environment variables.
If using Next.js, it must have a .eslintrc.json similar to plop-templates/example/.eslintrc.json
All Next.js examples should be using the same styling and layout provided by @vercel/examples-ui, its usage can be seen in the plop template

Adding a template

If you would like the example to be featured in vercel.com/templates then also add the front matter metadata to the top of the readme, like in bot-protection-datadome. To know all the possible values for each metadata take a look at internal/fields.json.

If you want to add related templates to your template, copy the slug from the other template into the relatedTemplates field, for example for vercel.com/templates/next.js/monorepo-turborepo the slug is monorepo-turborepo, as written in solutions/monorepo/README.md

The pre-commit hook

We use Husky to manage the pre-commit Git hook in this repo. Husky configures hooks automatically during install, so you don’t need to do anything special to get them working, but if it fails to install, you can run the following command to install it manually:

`1`	`pnpm run prepare`

Code changes automatically go through Prettier and ESLint when you make a commit, please do not skip these steps unless they’re broken and in that case let us known by creating an issue.

Read the Docs

If you have any questions or suggestions about the docs, feel free to open a discussion, or submit a PR with your suggestions!

Provide Feedback

Start a Discussion with a question, piece of feedback, or idea you want to share with the team.
Open an Issue if you believe you’ve encountered a bug that you want to flag for the team.

MiniCPM-V

Tue, 02 Sep 2025 15:29:41 +0800

OpenBMB/MiniCPM-V

A GPT-4o Level MLLM for Single Image, Multi Image and High-FPS Video Understanding on Your Phone

中文 | English

WeChat | Discord

MiniCPM-V 4.5 🤗 🤖 | MiniCPM-o 2.6 🤗 🤖 | 🍳 Cookbook | 📄 Technical Report (Coming Soon)

MiniCPM-V is a series of efficient end-side multimodal LLMs (MLLMs), which accept images, videos and text as inputs and deliver high-quality text outputs. MiniCPM-o additionally takes audio as inputs and provides high-quality speech outputs in an end-to-end fashion. Since February 2024, we have released 7 versions of the model, aiming to achieve strong performance and efficient deployment. The most notable models in the series currently include:

MiniCPM-V 4.5: 🔥🔥🔥 The latest and most capable model in the MiniCPM-V series. With a total of 8B parameters, this model outperforms GPT-4o-latest, Gemini-2.0 Pro, and Qwen2.5-VL 72B in vision-language capabilities, making it the most performant on-device multimodal model in the open-source community. This version brings new features including efficient high-FPS and long video understanding (up to 96x compression rate for video tokens), controllable hybrid fast/deep thinking, strong handwritten OCR and complex table/document parsing. It also advances MiniCPM-V’s popular features such as trustworthy behavior, multilingual support and end-side deployability.
MiniCPM-o 2.6: ⭐️⭐️⭐️ The most capable model in the MiniCPM-o series. With a total of 8B parameters, this end-to-end model achieves comparable performance to GPT-4o-202405 in vision, speech, and multimodal live streaming, making it one of the most versatile and performant models in the open-source community. For the new voice mode, MiniCPM-o 2.6 supports bilingual real-time speech conversation with configurable voices, and also allows for fun capabilities such as emotion/speed/style control, end-to-end voice cloning, role play, etc. Due to its superior token density, MiniCPM-o 2.6 can for the first time support multimodal live streaming on end-side devices such as iPad.

News

📌 Pinned

[2025.09.01] ⭐️⭐️⭐️ MiniCPM-V 4.5 has been officially supported by llama.cpp, vLLM, and LLaMA-Factory. You are welcome to use it directly through these official channels! Support for additional frameworks such as Ollama and SGLang is actively in progress.
[2025.08.26] 🔥🔥🔥 We open-source MiniCPM-V 4.5, which outperforms GPT-4o-latest, Gemini-2.0 Pro, and Qwen2.5-VL 72B. It advances popular capabilities of MiniCPM-V, and brings useful new features. Try it now!
[2025.08.01] ⭐️⭐️⭐️ We open-sourced the MiniCPM-V & o Cookbook! It provides comprehensive guides for diverse user scenarios, paired with our new Docs Site for smoother onboarding.
[2025.06.20] ⭐️⭐️⭐️ Our official Ollama repository is released. Try our latest models with one click！
[2025.03.01] 🚀🚀🚀 RLAIF-V, the alignment technique of MiniCPM-o, is accepted by CVPR 2025 Highlights！The code, dataset, paper are open-sourced!
[2025.01.24] 📢📢📢 MiniCPM-o 2.6 technical report is released! See here.
[2025.01.19] 📢 ATTENTION! We are currently working on merging MiniCPM-o 2.6 into the official repositories of llama.cpp, Ollama, and vllm. Until the merge is complete, please USE OUR LOCAL FORKS of llama.cpp, Ollama, and vllm. Using the official repositories before the merge may lead to unexpected issues.
[2025.01.19] ⭐️⭐️⭐️ MiniCPM-o tops GitHub Trending and reaches top-2 on Hugging Face Trending!
[2025.01.17] We have updated the usage of MiniCPM-o 2.6 int4 quantization version and resolved the model initialization error. Click here and try it now!
[2025.01.13] 🔥🔥🔥 We open-source MiniCPM-o 2.6, which matches GPT-4o-202405 on vision, speech and multimodal live streaming. It advances popular capabilities of MiniCPM-V 2.6, and supports various new fun features. Try it now!
[2024.08.17] 🚀🚀🚀 MiniCPM-V 2.6 is now fully supported by official llama.cpp! GGUF models of various sizes are available here.
[2024.08.06] 🔥🔥🔥 We open-source MiniCPM-V 2.6, which outperforms GPT-4V on single image, multi-image and video understanding. It advances popular features of MiniCPM-Llama3-V 2.5, and can support real-time video understanding on iPad. Try it now!
[2024.08.03] MiniCPM-Llama3-V 2.5 technical report is released! See here.
[2024.05.23] 🔥🔥🔥 MiniCPM-V tops GitHub Trending and Hugging Face Trending! Our demo, recommended by Hugging Face Gradio’s official account, is available here. Come and try it out!

Click to view more news.

[2025.08.02] 🚀🚀🚀 We open-source MiniCPM-V 4.0, which outperforms GPT-4.1-mini-20250414 in image understanding. It advances popular features of MiniCPM-V 2.6, and largely improves the efficiency. We also open-source the iOS App on iPhone and iPad. Try it now!
[2025.01.23] 💡💡💡 MiniCPM-o 2.6 is now supported by Align-Anything, a framework by PKU-Alignment Team for aligning any-to-any modality large models with human intentions. It supports DPO and SFT fine-tuning on both vision and audio. Try it now!
[2024.08.15] We now also support multi-image SFT. For more details, please refer to the document.
[2024.08.14] MiniCPM-V 2.6 now also supports fine-tuning with the SWIFT framework!
[2024.08.10] 🚀🚀🚀 MiniCPM-Llama3-V 2.5 is now fully supported by official llama.cpp! GGUF models of various sizes are available here.
[2024.07.19] MiniCPM-Llama3-V 2.5 supports vLLM now! See here.
[2024.06.03] Now, you can run MiniCPM-Llama3-V 2.5 on multiple low VRAM GPUs(12 GB or 16 GB) by distributing the model’s layers across multiple GPUs. For more details, check this link.
[2024.05.28] 🚀🚀🚀 MiniCPM-Llama3-V 2.5 now fully supports its feature in llama.cpp and Ollama! Please pull the latest code of our provided forks (llama.cpp, Ollama). GGUF models in various sizes are available here. MiniCPM-Llama3-V 2.5 series is not supported by the official repositories yet, and we are working hard to merge PRs. Please stay tuned!
[2024.05.28] 💫 We now support LoRA fine-tuning for MiniCPM-Llama3-V 2.5, using only 2 V100 GPUs! See more statistics here.
[2024.05.25] MiniCPM-Llama3-V 2.5 now supports streaming outputs and customized system prompts. Try it here!
[2024.05.24] We release the MiniCPM-Llama3-V 2.5 gguf, which supports llama.cpp inference and provides a 6~8 token/s smooth decoding on mobile phones. Try it now!
[2024.05.23] 🔍 We’ve released a comprehensive comparison between Phi-3-vision-128k-instruct and MiniCPM-Llama3-V 2.5, including benchmark evaluations, multilingual capabilities, and inference efficiency 🌟📊🌍🚀. Click here to view more details.
[2024.05.20] We open-soure MiniCPM-Llama3-V 2.5, it has improved OCR capability and supports 30+ languages, representing the first end-side MLLM achieving GPT-4V level performance! We provide efficient inference and simple fine-tuning. Try it now!
[2024.04.23] MiniCPM-V-2.0 supports vLLM now! Click here to view more details.
[2024.04.18] We create a HuggingFace Space to host the demo of MiniCPM-V 2.0 at here!
[2024.04.17] MiniCPM-V-2.0 supports deploying WebUI Demo now!
[2024.04.15] MiniCPM-V-2.0 now also supports fine-tuning with the SWIFT framework!
[2024.04.12] We open-source MiniCPM-V 2.0, which achieves comparable performance with Gemini Pro in understanding scene text and outperforms strong Qwen-VL-Chat 9.6B and Yi-VL 34B on OpenCompass, a comprehensive evaluation over 11 popular benchmarks. Click here to view the MiniCPM-V 2.0 technical blog.
[2024.03.14] MiniCPM-V now supports fine-tuning with the SWIFT framework. Thanks to Jintao for the contribution！
[2024.03.01] MiniCPM-V can now be deployed on Mac!
[2024.02.01] We open-source MiniCPM-V and OmniLMM-12B, which support efficient end-side deployment and powerful multimodal capabilities correspondingly.

MiniCPM-V 4.5

MiniCPM-V 4.5 is the latest and most capable model in the MiniCPM-V series. The model is built on Qwen3-8B and SigLIP2-400M with a total of 8B parameters. It exhibits a significant performance improvement over previous MiniCPM-V and MiniCPM-o models, and introduces new useful features. Notable features of MiniCPM-V 4.5 include:

🔥 State-of-the-art Vision-Language Capability. MiniCPM-V 4.5 achieves an average score of 77.0 on OpenCompass, a comprehensive evaluation of 8 popular benchmarks. With only 8B parameters, it surpasses widely used proprietary models like GPT-4o-latest, Gemini-2.0 Pro, and strong open-source models like Qwen2.5-VL 72B for vision-language capabilities, making it the most performant MLLM under 30B parameters.
🎬 Efficient High-FPS and Long Video Understanding. Powered by a new unified 3D-Resampler over images and videos, MiniCPM-V 4.5 can now achieve 96x compression rate for video tokens, where 6 448x448 video frames can be jointly compressed into 64 video tokens (normally 1,536 tokens for most MLLMs). This means that the model can perceive significantly more video frames without increasing the LLM inference cost. This brings state-of-the-art high-FPS (up to 10FPS) video understanding and long video understanding capabilities on Video-MME, LVBench, MLVU, MotionBench, FavorBench, etc., efficiently.
⚙️ Controllable Hybrid Fast/Deep Thinking. MiniCPM-V 4.5 supports both fast thinking for efficient frequent usage with competitive performance, and deep thinking for more complex problem solving. To cover efficiency and performance trade-offs in different user scenarios, this fast/deep thinking mode can be switched in a highly controlled fashion.
💪 Strong OCR, Document Parsing and Others. Based on LLaVA-UHD architecture, MiniCPM-V 4.5 can process high-resolution images with any aspect ratio and up to 1.8 million pixels (e.g., 1344x1344), using 4x fewer visual tokens than most MLLMs. The model achieves leading performance on OCRBench, surpassing proprietary models such as GPT-4o-latest and Gemini 2.5. It also achieves state-of-the-art performance for PDF document parsing capability on OmniDocBench among general MLLMs. Based on the latest RLAIF-V and VisCPM techniques, it features trustworthy behaviors, outperforming GPT-4o-latest on MMHal-Bench, and supports multilingual capabilities in more than 30 languages.
💫 Easy Usage. MiniCPM-V 4.5 can be easily used in various ways: (1) llama.cpp and ollama support for efficient CPU inference on local devices, (2) int4, GGUF and AWQ format quantized models in 16 sizes, (3) SGLang and vLLM support for high-throughput and memory-efficient inference, (4) fine-tuning on new domains and tasks with Transformers and LLaMA-Factory, (5) quick local WebUI demo, (6) optimized local iOS app on iPhone and iPad, and (7) online web demo on server. See our Cookbook for full usage!

Key Techniques

Architechture: Unified 3D-Resampler for High-density Video Compression. MiniCPM-V 4.5 introduces a 3D-Resampler that overcomes the performance-efficiency trade-off in video understanding. By grouping and jointly compressing up to 6 consecutive video frames into just 64 tokens (the same token count used for a single image in MiniCPM-V series), MiniCPM-V 4.5 achieves a 96× compression rate for video tokens. This allows the model to process more video frames without additional LLM computational cost, enabling high-FPS video and long video understanding. The architecture supports unified encoding for images, multi-image inputs, and videos, ensuring seamless capability and knowledge transfer.
Pre-training: Unified Learning for OCR and Knowledge from Documents. Existing MLLMs learn OCR capability and knowledge from documents in isolated training approaches. We observe that the essential difference between these two training approaches is the visibility of the text in images. By dynamically corrupting text regions in documents with varying noise levels and asking the model to reconstruct the text, the model learns to adaptively and properly switch between accurate text recognition (when text is visible) and multimodal context-based knowledge reasoning (when text is heavily obscured). This eliminates reliance on error-prone document parsers in knowledge learning from documents, and prevents hallucinations from over-augmented OCR data, resulting in top-tier OCR and multimodal knowledge performance with minimal engineering overhead.
Post-training: Hybrid Fast/Deep Thinking with Multimodal RL. MiniCPM-V 4.5 offers a balanced reasoning experience through two switchable modes: fast thinking for efficient daily use and deep thinking for complex tasks. Using a new hybrid reinforcement learning method, the model jointly optimizes both modes, significantly enhancing fast-mode performance without compromising deep-mode capability. Incorporated with RLPR and RLAIF-V, it generalizes robust reasoning skills from broad multimodal data while effectively reducing hallucinations.

Evaluation

Inference Efficiency

OpenCompass

Model	Size	Avg Score ↑	Total Inference Time ↓
GLM-4.1V-9B-Thinking	10.3B	76.6	17.5h
MiMo-VL-7B-RL	8.3B	76.4	11h
MiniCPM-V 4.5	8.7B	77.0	7.5h

Video-MME

Model	Size	Avg Score ↑	Total Inference Time ↓	GPU Mem ↓
Qwen2.5-VL-7B-Instruct	8.3B	71.6	3h	60G
GLM-4.1V-9B-Thinking	10.3B	73.6	2.63h	32G
MiniCPM-V 4.5	8.7B	73.5	0.26h	28G

Both Video-MME and OpenCompass were evaluated using 8×A100 GPUs for inference. The reported inference time of Video-MME includes full model-side computation, and excludes the external cost of video frame extraction (dependent on specific frame extraction tools) for fair comparison.

Examples

Click to view more cases.

We deploy MiniCPM-V 4.5 on iPad M4 with iOS demo. The demo video is the raw screen recording without edition.

MiniCPM-o 2.6

MiniCPM-o 2.6 is the latest and most capable model in the MiniCPM-o series. The model is built in an end-to-end fashion based on SigLip-400M, Whisper-medium-300M, ChatTTS-200M, and Qwen2.5-7B with a total of 8B parameters. It exhibits a significant performance improvement over MiniCPM-V 2.6, and introduces new features for real-time speech conversation and multimodal live streaming. Notable features of MiniCPM-o 2.6 include:

🔥 Leading Visual Capability. MiniCPM-o 2.6 achieves an average score of 70.2 on OpenCompass, a comprehensive evaluation of 8 popular benchmarks. With only 8B parameters, it surpasses widely used proprietary models like GPT-4o-202405, Gemini 1.5 Pro, and Claude 3.5 Sonnet for single image understanding. It also outperforms GPT-4V and Claude 3.5 Sonnet in multi-image and video understanding, and shows promising in-context learning capability.
🎙 State-of-the-art Speech Capability. MiniCPM-o 2.6 supports bilingual real-time speech conversation with configurable voices in English and Chinese. It outperforms GPT-4o-realtime on audio understanding tasks such as ASR and STT translation, and shows state-of-the-art performance on speech conversation in both semantic and acoustic evaluations in the open-source community. It also allows for fun features such as emotion/speed/style control, end-to-end voice cloning, role play, etc.
🎬 Strong Multimodal Live Streaming Capability. As a new feature, MiniCPM-o 2.6 can accept continuous video and audio streams independent of user queries, and support real-time speech interaction. It outperforms GPT-4o-202408 and Claude 3.5 Sonnet and shows state-of-the-art performance in the open-source community on StreamingBench, a comprehensive benchmark for real-time video understanding, omni-source (video & audio) understanding, and multimodal contextual understanding.
💪 Strong OCR Capability and Others. Advancing popular visual capabilities from MiniCPM-V series, MiniCPM-o 2.6 can process images with any aspect ratio and up to 1.8 million pixels (e.g., 1344x1344). It achieves state-of-the-art performance on OCRBench for models under 25B, surpassing proprietary models such as GPT-4o-202405. Based on the latest RLAIF-V and VisCPM techniques, it features trustworthy behaviors, outperforming GPT-4o and Claude 3.5 Sonnet on MMHal-Bench, and supports multilingual capabilities on more than 30 languages.
🚀 Superior Efficiency. In addition to its friendly size, MiniCPM-o 2.6 also shows state-of-the-art token density (i.e., the number of pixels encoded into each visual token). It produces only 640 tokens when processing a 1.8M pixel image, which is 75% fewer than most models. This directly improves the inference speed, first-token latency, memory usage, and power consumption. As a result, MiniCPM-o 2.6 can efficiently support multimodal live streaming on end-side devices such as iPads.
💫 Easy Usage. MiniCPM-o 2.6 can be easily used in various ways: (1) llama.cpp support for efficient CPU inference on local devices, (2) int4 and GGUF format quantized models in 16 sizes, (3) vLLM support for high-throughput and memory-efficient inference, (4) fine-tuning on new domains and tasks with LLaMA-Factory, (5) quick local WebUI demo, and (6) online web demo on server.

Model Architecture.

End-to-end Omni-modal Architecture. Different modality encoders/decoders are connected and trained in an end-to-end fashion to fully exploit rich multimodal knowledge. The model is trained in a fully end-to-end manner with only CE loss.
Omni-modal Live Streaming Mechanism. (1) We change the offline modality encoder/decoders into online ones for streaming inputs/outputs. (2) We devise a time-division multiplexing (TDM) mechanism for omni-modality streaming processing in the LLM backbone. It divides parallel omni-modality streams into sequential info within small periodic time slices.
Configurable Speech Modeling Design. We devise a multimodal system prompt, including traditional text system prompt, and a new audio system prompt to determine the assistant voice. This enables flexible voice configurations in inference time, and also facilitates end-to-end voice cloning and description-based voice creation.

Evaluation

Click to view visual understanding results.

Image Understanding

Model	Size	Token Density⁺	OpenCompass	OCRBench	MathVista mini	ChartQA	MMVet	MMStar	MME	MMB1.1 test	AI2D	MMMU val	HallusionBench	TextVQA val	DocVQA test	MathVerse mini	MathVision	MMHal Score
Proprietary
GPT-4o-20240513	-	1088	69.9	736	61.3	85.7	69.1	63.9	2328.7	82.2	84.6	69.2	55.0	-	92.8	50.2	30.4	3.6
Claude3.5-Sonnet	-	750	67.9	788	61.6	90.8	66.0	62.2	1920.0	78.5	80.2	65.9	49.9	-	95.2	-	-	3.4
Gemini 1.5 Pro	-	-	64.4	754	57.7	81.3	64.0	59.1	2110.6	73.9	79.1	60.6	45.6	73.5	86.5	-	19.2	-
GPT-4o-mini-20240718	-	1088	64.1	785	52.4	-	66.9	54.8	2003.4	76.0	77.8	60.0	46.1	-	-	-	-	3.3
Open Source
Cambrian-34B	34B	1820	58.3	591	50.3	75.6	53.2	54.2	2049.9	77.8	79.5	50.4	41.6	76.7	75.5	-	-	-
GLM-4V-9B	13B	784	59.1	776	51.1	-	58.0	54.8	2018.8	67.9	71.2	46.9	45.0	-	-	-	-	-
Pixtral-12B	12B	256	61.0	685	56.9	81.8	58.5	54.5	-	72.7	79.0	51.1	47.0	75.7	90.7	-	-	-
VITA-1.5	8B	784	63.3	741	66.2	-	52.7	60.2	2328.1	76.8	79.2	52.6	44.6	-	-	-	-	-
DeepSeek-VL2-27B (4B)	27B	672	66.4	809	63.9	86.0	60.0	61.9	2253.0	81.2	83.8	54.0	45.3	84.2	93.3	-	-	3.0
Qwen2-VL-7B	8B	784	67.1	866	58.2	83.0	62.0	60.7	2326.0	81.8	83.0	54.1	50.6	84.3	94.5	31.9	16.3	3.2
LLaVA-OneVision-72B	72B	182	68.1	741	67.5	83.7	60.6	65.8	2261.0	85.0	85.6	56.8	49.0	80.5	91.3	39.1	-	3.5
InternVL2.5-8B	8B	706	68.3	822	64.4	84.8	62.8	62.8	2344.0	83.6	84.5	56.0	50.1	79.1	93.0	39.5	19.7	3.4
MiniCPM-V 2.6	8B	2822	65.2	852*	60.6	79.4	60.0	57.5	2348.4*	78.0	82.1	49.8*	48.1*	80.1	90.8	25.7	18.3	3.6
MiniCPM-o 2.6	8B	2822	70.2	897*	71.9*	86.9*	67.5	64.0	2372.0*	80.5	85.8	50.4*	51.9	82.0	93.5	41.4*	23.1*	3.8

* We evaluate this benchmark using chain-of-thought prompting. Specifically, for MME, we used this technique only for the Cognition set.

⁺ Token Density: number of pixels encoded into each visual token at maximum resolution, i.e., # pixels at maximum resolution / # visual tokens.

Note: For proprietary models, we calculate token density based on the image encoding charging strategy defined in the official API documentation, which provides an upper-bound estimation.

Multi-image and Video Understanding

Model	Size	BLINK val	Mantis Eval	MIRB	Video-MME (wo / w subs)
Proprietary
GPT-4o-20240513	-	68.0	-	-	71.9/77.2
GPT4V	-	54.6	62.7	53.1	59.9/63.3
Open-source
VITA-1.5	8B	45.0	-	-	56.1/58.7
LLaVA-NeXT-Interleave 14B	14B	52.6	66.4	30.2	-
LLaVA-OneVision-72B	72B	55.4	77.6	-	66.2/69.5
MANTIS 8B	8B	49.1	59.5	34.8	-
Qwen2-VL-7B	8B	53.2	69.6*	67.6*	63.3/69.0
InternVL2.5-8B	8B	54.8	67.7	52.5	64.2/66.9
MiniCPM-V 2.6	8B	53.0	69.1	53.8	60.9/63.6
MiniCPM-o 2.6	8B	56.7	71.9	58.6	63.9/67.9

* We evaluate officially released checkpoints by ourselves.

Click to view audio understanding and speech conversation results.

Audio Understanding

Task	Size	ASR (zh)	ASR (en)	AST	Emotion
Metric		CER↓	WER↓	BLEU↑	ACC↑
Dataset		AISHELL-1	Fleurs zh	WenetSpeech test-net	LibriSpeech test-clean	GigaSpeech	TED-LIUM	CoVoST en2zh	CoVoST zh2en	MELD emotion
Proprietary
GPT-4o-Realtime	-	7.3*	5.4*	28.9*	2.6*	12.9*	4.8*	37.1*	15.7*	33.2*
Gemini 1.5 Pro	-	4.5*	5.9*	14.3*	2.9*	10.6*	3.0*	47.3*	22.6*	48.4*
Open-Source
Qwen2-Audio-7B	8B	-	7.5	-	1.6	-	-	45.2	24.4	55.3
Qwen2-Audio-7B-Instruct	8B	2.6*	6.9*	10.3*	3.1*	9.7*	5.9*	39.5*	22.9*	17.4*
VITA-1.5	8B	2.16	-	8.4	3.4	-	-	-	-	-
GLM-4-Voice-Base	9B	2.5	-	-	2.8	-	-	-	-
MiniCPM-o 2.6	8B	1.6	4.4	6.9	1.7	8.7	3.0	48.2	27.2	52.4

* We evaluate officially released checkpoints by ourselves.

Speech Generation

Task	Size	SpeechQA
Metric		ACC↑	G-Eval (10 point)↑	Semantic ELO score↑	Acoustic ELO score↑	Overall ELO score↑	UTMOS↑	ASR-WER↓
Dataset		Speech Llama Q.	Speech Web Q.	Speech Trivia QA	Speech AlpacaEval	AudioArena
Proprietary
GPT-4o-Realtime		71.7	51.6	69.7	7.4	1157	1203	1200	4.2	2.3
Open-Source
GLM-4-Voice	9B	50.0	32.0	36.4	5.1	999	1147	1035	4.1	11.7
Llama-Omni	8B	45.3	22.9	10.7	3.9	960	878	897	3.2	24.3
VITA-1.5	8B	46.7	28.1	23.3	2.0	-	-	-	-	-
Moshi	7B	43.7	23.8	16.7	2.4	871	808	875	2.8	8.2
Mini-Omni	1B	22.0	12.8	6.9	2.5	926	803	865	3.4	10.0
MiniCPM-o 2.6	8B	61.0	40.0	40.2	5.1	1088	1163	1131	4.2	9.8

All results are from AudioEvals, and the evaluation methods along with further details can be found in AudioEvals.

End-to-end Voice Cloning

Task	Voice cloning
Metric	SIMO↑	SIMO↑
Dataset	Seed-TTS test-zh	Seed-TTS test-en
F5-TTS	76	67
CosyVoice	75	64
FireRedTTS	63	46
MiniCPM-o 2.6	57	47

Click to view multimodal live streaming results.

Multimodal Live Streaming: results on StreamingBench

Model	Size	Real-Time Video Understanding	Omni-Source Understanding	Contextual Understanding	Overall
Proprietary
Gemini 1.5 Pro	-	77.4	67.8	51.1	70.3
GPT-4o-202408	-	74.5	51.0	48.0	64.1
Claude-3.5-Sonnet	-	74.0	41.4	37.8	59.7
Open-source
VILA-1.5	8B	61.5	37.5	26.7	49.5
LongVA	7B	63.1	35.9	30.2	50.7
LLaVA-Next-Video-34B	34B	69.8	41.7	34.3	56.7
Qwen2-VL-7B	8B	71.2	40.7	33.1	57.0
InternVL2-8B	8B	70.1	42.7	34.1	57.0
VITA-1.5	8B	70.9	40.8	35.8	57.4
LLaVA-OneVision-7B	8B	74.3	40.8	31.0	58.4
InternLM-XC2.5-OL-7B	8B	75.4	46.2	33.6	60.8
MiniCPM-V 2.6	8B	72.4	40.2	33.4	57.7
MiniCPM-o 2.6	8B	79.9	53.4	38.5	66.0

Examples

We deploy MiniCPM-o 2.6 on end devices. The demo video is the raw-speed recording on an iPad Pro and a Web demo.

Legacy Models

Model	Introduction and Guidance
MiniCPM-V 4.0	Document
MiniCPM-V 2.6	Document
MiniCPM-Llama3-V 2.5	Document
MiniCPM-V 2.0	Document
MiniCPM-V 1.0	Document
OmniLMM-12B	Document

MiniCPM-V & o Cookbook

Discover comprehensive, ready-to-deploy solutions for the MiniCPM-V and MiniCPM-o model series in our structured cookbook, which empowers developers to rapidly implement multimodal AI applications with integrated vision, speech, and live-streaming capabilities. Key features include:

Easy Usage Documentation

Our comprehensive documentation website presents every recipe in a clear, well-organized manner. All features are displayed at a glance, making it easy for you to quickly find exactly what you need.

Broad User Spectrum

We support a wide range of users, from individuals to enterprises and researchers.

Individuals: Enjoy effortless inference using Ollama and Llama.cpp with minimal setup.
Enterprises: Achieve high-throughput, scalable performance with vLLM and SGLang.
Researchers: Leverage advanced frameworks including Transformers, LLaMA-Factory, SWIFT, and Align-anything to enable flexible model development and cutting-edge experimentation.

Versatile Deployment Scenarios

Our ecosystem delivers optimal solution for a variety of hardware environments and deployment demands.

Web demo: Launch interactive multimodal AI web demo with FastAPI.
Quantized deployment: Maximize efficiency and minimize resource consumption using GGUF and BNB.
End devices: Bring powerful AI experiences to iPhone and iPad, supporting offline and privacy-sensitive applications.

Chat with Our Demo on Gradio 🤗

We provide online and local demos powered by Hugging Face Gradio , the most popular model deployment framework nowadays. It supports streaming outputs, progress bars, queuing, alerts, and other useful features.

Online Demo

Click here to try out the online demo of MiniCPM-o 2.6 | MiniCPM-V 2.6 | MiniCPM-Llama3-V 2.5 | MiniCPM-V 2.0.

Local WebUI Demo

You can easily build your own local WebUI demo using the following commands.

Please ensure that transformers==4.44.2 is installed, as other versions may have compatibility issues.

If you are using an older version of PyTorch, you might encounter this issue "weight_norm_fwd_first_dim_kernel" not implemented for 'BFloat16', Please add self.minicpmo_model.tts.float() during the model initialization.

For real-time voice/video call demo:

launch model server:

1
2
3

pip install -r requirements_o2.6.txt

python web_demos/minicpm-o_2.6/model_server.py

launch web server:

# Make sure Node and PNPM is installed.
sudo apt-get update
sudo apt-get install nodejs npm
npm install -g pnpm


cd web_demos/minicpm-o_2.6/web_server
# create ssl cert for https, https is required to request camera and microphone permissions.
bash ./make_ssl_cert.sh  # output key.pem and cert.pem

pnpm install  # install requirements
pnpm run dev  # start server

Open https://localhost:8088/ in browser and enjoy the real-time voice/video call.

For chatbot demo:

1
2
3

pip install -r requirements_o2.6.txt

python web_demos/minicpm-o_2.6/chatbot_web_demo_o2.6.py

Open http://localhost:8000/ in browser and enjoy the vision mode chatbot.

Inference

Model Zoo

Model	Device	Memory	Description	Download
MiniCPM-V 4.5	GPU	18 GB	The latest version, strong end-side multimodal performance for single image, multi-image and video understanding.	🤗
MiniCPM-V 4.5 gguf	CPU	8 GB	The gguf version, lower memory usage and faster inference.	🤗
MiniCPM-V 4.5 int4	GPU	9 GB	The int4 quantized version, lower GPU memory usage.	🤗
MiniCPM-V 4.5 AWQ	GPU	9 GB	The int4 quantized version, lower GPU memory usage.	🤗
MiniCPM-o 2.6	GPU	18 GB	The latest version, achieving GPT-4o level performance for vision, speech and multimodal live streaming on end-side devices.	🤗
MiniCPM-o 2.6 gguf	CPU	8 GB	The gguf version, lower memory usage and faster inference.	🤗
MiniCPM-o 2.6 int4	GPU	9 GB	The int4 quantized version, lower GPU memory usage.	🤗

Multi-turn Conversation

If you wish to enable long-thinking mode, provide the argument enable_thinking=True to the chat function.

`1`	`pip install -r requirements_o2.6.txt`

Please refer to the following codes to run.

import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer

torch.manual_seed(100)

model = AutoModel.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True, # or openbmb/MiniCPM-o-2_6
    attn_implementation='sdpa', torch_dtype=torch.bfloat16) # sdpa or flash_attention_2, no eager
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True) # or openbmb/MiniCPM-o-2_6

image = Image.open('./assets/minicpmo2_6/show_demo.jpg').convert('RGB')

enable_thinking=False # If `enable_thinking=True`, the long-thinking mode is enabled.

# First round chat 
question = "What is the landform in the picture?"
msgs = [{'role': 'user', 'content': [image, question]}]

answer = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    enable_thinking=enable_thinking
)
print(answer)

# Second round chat, pass history context of multi-turn conversation
msgs.append({"role": "assistant", "content": [answer]})
msgs.append({"role": "user", "content": ["What should I pay attention to when traveling here?"]})

answer = model.chat(
    msgs=msgs,
    tokenizer=tokenizer
)
print(answer)

You will get the following output:

# round1
The landform in the picture is karst topography. Karst landscapes are characterized by distinctive, jagged limestone hills or mountains with steep, irregular peaks and deep valleys—exactly what you see here These unique formations result from the dissolution of soluble rocks like limestone over millions of years through water erosion.

This scene closely resembles the famous karst landscape of Guilin and Yangshuo in China’s Guangxi Province. The area features dramatic, pointed limestone peaks rising dramatically above serene rivers and lush green forests, creating a breathtaking and iconic natural beauty that attracts millions of visitors each year for its picturesque views.

# round2
When traveling to a karst landscape like this, here are some important tips:

1. Wear comfortable shoes: The terrain can be uneven and hilly.
2. Bring water and snacks for energy during hikes or boat rides.
3. Protect yourself from the sun with sunscreen, hats, and sunglasses—especially since you’ll likely spend time outdoors exploring scenic spots.
4. Respect local customs and nature regulations by not littering or disturbing wildlife.

By following these guidelines, you'll have a safe and enjoyable trip while appreciating the stunning natural beauty of places such as Guilin’s karst mountains.

Chat with Multiple Images

Click to view Python code running MiniCPM-V-4_5 with multiple images input.

import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True,  # or openbmb/MiniCPM-o-2_6
    attn_implementation='sdpa', torch_dtype=torch.bfloat16) # sdpa or flash_attention_2, no eager
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True)  # or openbmb/MiniCPM-o-2_6

image1 = Image.open('image1.jpg').convert('RGB')
image2 = Image.open('image2.jpg').convert('RGB')
question = 'Compare image 1 and image 2, tell me about the differences between image 1 and image 2.'

msgs = [{'role': 'user', 'content': [image1, image2, question]}]

answer = model.chat(
    msgs=msgs,
    tokenizer=tokenizer
)
print(answer)

In-context Few-shot Learning

Click to view Python code running MiniCPM-V-4_5 with few-shot input.

import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True,  # or openbmb/MiniCPM-o-2_6
    attn_implementation='sdpa', torch_dtype=torch.bfloat16) # sdpa or flash_attention_2, no eager
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True)  # or openbmb/MiniCPM-o-2_6

question = "production date" 
image1 = Image.open('example1.jpg').convert('RGB')
answer1 = "2023.08.04"
image2 = Image.open('example2.jpg').convert('RGB')
answer2 = "2007.04.24"
image_test = Image.open('test.jpg').convert('RGB')

msgs = [
    {'role': 'user', 'content': [image1, question]}, {'role': 'assistant', 'content': [answer1]},
    {'role': 'user', 'content': [image2, question]}, {'role': 'assistant', 'content': [answer2]},
    {'role': 'user', 'content': [image_test, question]}
]

answer = model.chat(
    msgs=msgs,
    tokenizer=tokenizer
)
print(answer)

Chat with Video

Click to view Python code running MiniCPM-V-4_5 by with video input and 3D-Resampler.

## The 3d-resampler compresses multiple frames into 64 tokens by introducing temporal_ids. 
# To achieve this, you need to organize your video data into two corresponding sequences: 
#   frames: List[Image]
#   temporal_ids: List[List[Int]].

import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer
from decord import VideoReader, cpu    # pip install decord
from scipy.spatial import cKDTree
import numpy as np
import math

model = AutoModel.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True,  # or openbmb/MiniCPM-o-2_6
    attn_implementation='sdpa', torch_dtype=torch.bfloat16) # sdpa or flash_attention_2, no eager
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True)  # or openbmb/MiniCPM-o-2_6

MAX_NUM_FRAMES=180 # Indicates the maximum number of frames received after the videos are packed. The actual maximum number of valid frames is MAX_NUM_FRAMES * MAX_NUM_PACKING.
MAX_NUM_PACKING=3  # indicates the maximum packing number of video frames. valid range: 1-6
TIME_SCALE = 0.1 

def map_to_nearest_scale(values, scale):
    tree = cKDTree(np.asarray(scale)[:, None])
    _, indices = tree.query(np.asarray(values)[:, None])
    return np.asarray(scale)[indices]


def group_array(arr, size):
    return [arr[i:i+size] for i in range(0, len(arr), size)]

def encode_video(video_path, choose_fps=3, force_packing=None):
    def uniform_sample(l, n):
        gap = len(l) / n
        idxs = [int(i * gap + gap / 2) for i in range(n)]
        return [l[i] for i in idxs]
    vr = VideoReader(video_path, ctx=cpu(0))
    fps = vr.get_avg_fps()
    video_duration = len(vr) / fps
        
    if choose_fps * int(video_duration) <= MAX_NUM_FRAMES:
        packing_nums = 1
        choose_frames = round(min(choose_fps, round(fps)) * min(MAX_NUM_FRAMES, video_duration))
        
    else:
        packing_nums = math.ceil(video_duration * choose_fps / MAX_NUM_FRAMES)
        if packing_nums <= MAX_NUM_PACKING:
            choose_frames = round(video_duration * choose_fps)
        else:
            choose_frames = round(MAX_NUM_FRAMES * MAX_NUM_PACKING)
            packing_nums = MAX_NUM_PACKING

    frame_idx = [i for i in range(0, len(vr))]      
    frame_idx =  np.array(uniform_sample(frame_idx, choose_frames))

    if force_packing:
        packing_nums = min(force_packing, MAX_NUM_PACKING)
    
    print(video_path, ' duration:', video_duration)
    print(f'get video frames={len(frame_idx)}, packing_nums={packing_nums}')
    
    frames = vr.get_batch(frame_idx).asnumpy()

    frame_idx_ts = frame_idx / fps
    scale = np.arange(0, video_duration, TIME_SCALE)

    frame_ts_id = map_to_nearest_scale(frame_idx_ts, scale) / TIME_SCALE
    frame_ts_id = frame_ts_id.astype(np.int32)

    assert len(frames) == len(frame_ts_id)

    frames = [Image.fromarray(v.astype('uint8')).convert('RGB') for v in frames]
    frame_ts_id_group = group_array(frame_ts_id, packing_nums)
    
    return frames, frame_ts_id_group


video_path="video_test.mp4"
fps = 5 # fps for video
force_packing = None # You can set force_packing to ensure that 3D packing is forcibly enabled; otherwise, encode_video will dynamically set the packing quantity based on the duration.
frames, frame_ts_id_group = encode_video(video_path, fps, force_packing=force_packing)

question = "Describe the video"
msgs = [
    {'role': 'user', 'content': frames + [question]}, 
]


answer = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    use_image_id=False,
    max_slice_nums=1,
    temporal_ids=frame_ts_id_group
)
print(answer)

Speech and Audio Mode

Model initialization

import torch
import librosa
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained('openbmb/MiniCPM-o-2_6', trust_remote_code=True,
    attn_implementation='sdpa', torch_dtype=torch.bfloat16) # sdpa or flash_attention_2, no eager
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-o-2_6', trust_remote_code=True)

model.init_tts()
model.tts.float()

Mimick

Mimick task reflects a model’s end-to-end speech modeling capability. The model takes audio input, and outputs an ASR transcription and subsequently reconstructs the original audio with high similarity. The higher the similarity between the reconstructed audio and the original audio, the stronger the model’s foundational capability in end-to-end speech modeling.

mimick_prompt = "Please repeat each user's speech, including voice style and speech content."
audio_input, _ = librosa.load('./assets/input_examples/Trump_WEF_2018_10s.mp3', sr=16000, mono=True) # load the audio to be mimicked

# `./assets/input_examples/fast-pace.wav`, 
# `./assets/input_examples/chi-english-1.wav` 
# `./assets/input_examples/exciting-emotion.wav` 
# for different aspects of speech-centric features.

msgs = [{'role': 'user', 'content': [mimick_prompt, audio_input]}]
res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    temperature=0.3,
    generate_audio=True,
    output_audio_path='output_mimick.wav', # save the tts result to output_audio_path
)

General Speech Conversation with Configurable Voices

A general usage scenario of MiniCPM-o-2.6 is role-playing a specific character based on the audio prompt. It will mimic the voice of the character to some extent and act like the character in text, including language style. In this mode, MiniCPM-o-2.6 sounds more natural and human-like. Self-defined audio prompts can be used to customize the voice of the character in an end-to-end manner.

ref_audio, _ = librosa.load('./assets/input_examples/icl_20.wav', sr=16000, mono=True) # load the reference audio
sys_prompt = model.get_sys_prompt(ref_audio=ref_audio, mode='audio_roleplay', language='en')

# round one
user_question = {'role': 'user', 'content': [librosa.load('xxx.wav', sr=16000, mono=True)[0]]}
msgs = [sys_prompt, user_question]
res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_roleplay_round_1.wav',
)

# round two
history = msgs.append({'role': 'assistant', 'content': res})
user_question = {'role': 'user', 'content': [librosa.load('xxx.wav', sr=16000, mono=True)[0]]}
msgs = history.append(user_question)
res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_roleplay_round_2.wav',
)
print(res)

Speech Conversation as an AI Assistant

An enhanced feature of MiniCPM-o-2.6 is to act as an AI assistant, but only with limited choice of voices. In this mode, MiniCPM-o-2.6 is less human-like and more like a voice assistant. In this mode, the model is more instruction-following. For demo, you are suggested to use assistant_female_voice, assistant_male_voice, and assistant_default_female_voice. Other voices may work but not as stable as the default voices.

Please note that, assistant_female_voice and assistant_male_voice are more stable but sounds like robots, while assistant_default_female_voice is more human-alike but not stable, its voice often changes in multiple turns. We suggest you to try stable voices assistant_female_voice and assistant_male_voice.

ref_audio, _ = librosa.load('./assets/input_examples/assistant_female_voice.wav', sr=16000, mono=True) # or use `./assets/input_examples/assistant_male_voice.wav`
sys_prompt = model.get_sys_prompt(ref_audio=ref_audio, mode='audio_assistant', language='en') 
user_question = {'role': 'user', 'content': [librosa.load('xxx.wav', sr=16000, mono=True)[0]]} # load the user's audio question

# round one
msgs = [sys_prompt, user_question]
res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_assistant_round_1.wav',
)

# round two
history = msgs.append({'role': 'assistant', 'content': res})
user_question = {'role': 'user', 'content': [librosa.load('xxx.wav', sr=16000, mono=True)[0]]}
msgs = history.append(user_question)
res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_assistant_round_2.wav',
)
print(res)

Instruction-to-Speech

MiniCPM-o-2.6 can also do Instruction-to-Speech, aka Voice Creation. You can describe a voice in detail, and the model will generate a voice that matches the description. For more Instruction-to-Speech sample instructions, you can refer to https://voxinstruct.github.io/VoxInstruct/.

instruction = 'Speak like a male charming superstar, radiating confidence and style in every word.'

msgs = [{'role': 'user', 'content': [instruction]}]

res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_voice_creation.wav',
)

Voice Cloning

MiniCPM-o-2.6 can also do zero-shot text-to-speech, aka Voice Cloning. With this mode, model will act like a TTS model.

ref_audio, _ = librosa.load('./assets/input_examples/icl_20.wav', sr=16000, mono=True) # load the reference audio
sys_prompt = model.get_sys_prompt(ref_audio=ref_audio, mode='voice_cloning', language='en')
text_prompt = f"Please read the text below."
user_question = {'role': 'user', 'content': [text_prompt, "content that you want to read"]}

msgs = [sys_prompt, user_question]
res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_voice_cloning.wav',
)

Addressing Various Audio Understanding Tasks

MiniCPM-o-2.6 can also be used to address various audio understanding tasks, such as ASR, speaker analysis, general audio captioning, and sound scene tagging.

For audio-to-text tasks, you can use the following prompts:

ASR with ZH(same as AST en2zh): 请仔细听这段音频片段，并将其内容逐字记录。
ASR with EN(same as AST zh2en): Please listen to the audio snippet carefully and transcribe the content.
Speaker Analysis: Based on the speaker's content, speculate on their gender, condition, age range, and health status.
General Audio Caption: Summarize the main content of the audio.
General Sound Scene Tagging: Utilize one keyword to convey the audio's content or the associated scene.

task_prompt = "Please listen to the audio snippet carefully and transcribe the content." + "\n" # can change to other prompts.
audio_input, _ = librosa.load('./assets/input_examples/audio_understanding.mp3', sr=16000, mono=True) # load the audio to be captioned

msgs = [{'role': 'user', 'content': [task_prompt, audio_input]}]

res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    max_new_tokens=128,
    use_tts_template=True,
    generate_audio=True,
    temperature=0.3,
    output_audio_path='result_audio_understanding.wav',
)
print(res)

Multimodal Live Streaming

Click to view Python code running MiniCPM-o 2.6 with chat inference.

import math
import numpy as np
from PIL import Image
from moviepy.editor import VideoFileClip
import tempfile
import librosa
import soundfile as sf
import torch
from transformers import AutoModel, AutoTokenizer

def get_video_chunk_content(video_path, flatten=True):
    video = VideoFileClip(video_path)
    print('video_duration:', video.duration)
    
    with tempfile.NamedTemporaryFile(suffix=".wav", delete=True) as temp_audio_file:
        temp_audio_file_path = temp_audio_file.name
        video.audio.write_audiofile(temp_audio_file_path, codec="pcm_s16le", fps=16000)
        audio_np, sr = librosa.load(temp_audio_file_path, sr=16000, mono=True)
    num_units = math.ceil(video.duration)
    
    # 1 frame + 1s audio chunk
    contents= []
    for i in range(num_units):
        frame = video.get_frame(i+1)
        image = Image.fromarray((frame).astype(np.uint8))
        audio = audio_np[sr*i:sr*(i+1)]
        if flatten:
            contents.extend(["<unit>", image, audio])
        else:
            contents.append(["<unit>", image, audio])
    
    return contents


model = AutoModel.from_pretrained('openbmb/MiniCPM-o-2_6', trust_remote_code=True,
    attn_implementation='sdpa', torch_dtype=torch.bfloat16)
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-o-2_6', trust_remote_code=True)

model.init_tts()

# If you are using an older version of PyTorch, you might encounter this issue "weight_norm_fwd_first_dim_kernel" not implemented for 'BFloat16', Please convert the TTS to float32 type.
# model.tts.float()

# https://huggingface.co/openbmb/MiniCPM-o-2_6/blob/main/assets/Skiing.mp4
video_path="assets/Skiing.mp4"
sys_msg = model.get_sys_prompt(mode='omni', language='en')
# if use voice clone prompt, please set ref_audio
# ref_audio_path = '/path/to/ref_audio'
# ref_audio, _ = librosa.load(ref_audio_path, sr=16000, mono=True)
# sys_msg = model.get_sys_prompt(ref_audio=ref_audio, mode='omni', language='en')

contents = get_video_chunk_content(video_path)
msg = {"role":"user", "content": contents}
msgs = [sys_msg, msg]

# please set generate_audio=True and output_audio_path to save the tts result
generate_audio = True
output_audio_path = 'output.wav'

res = model.chat(
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    temperature=0.5,
    max_new_tokens=4096,
    omni_input=True, # please set omni_input=True when omni inference
    use_tts_template=True,
    generate_audio=generate_audio,
    output_audio_path=output_audio_path,
    max_slice_nums=1,
    use_image_id=False,
    return_dict=True
)
print(res)

Click to view Python code running MiniCPM-o 2.6 with streaming inference.

Note: The streaming inference has a slight performance degradation because the audio encoding is not global.

# a new conversation need reset session first, it will reset the kv-cache
model.reset_session()

contents = get_video_chunk_content(video_path, flatten=False)
session_id = '123'
generate_audio = True

# 1. prefill system prompt
res = model.streaming_prefill(
    session_id=session_id,
    msgs=[sys_msg], 
    tokenizer=tokenizer
)

# 2. prefill video/audio chunks
for content in contents:
    msgs = [{"role":"user", "content": content}]
    res = model.streaming_prefill(
        session_id=session_id,
        msgs=msgs, 
        tokenizer=tokenizer
    )

# 3. generate
res = model.streaming_generate(
    session_id=session_id,
    tokenizer=tokenizer,
    temperature=0.5,
    generate_audio=generate_audio
)

audios = []
text = ""

if generate_audio:
    for r in res:
        audio_wav = r.audio_wav
        sampling_rate = r.sampling_rate
        txt = r.text

        audios.append(audio_wav)
        text += txt
        
    res = np.concatenate(audios)
    sf.write("output.wav", res, samplerate=sampling_rate)
    print("text:", text)
    print("audio saved to output.wav")
else:
    for r in res:
        text += r['text']
    print("text:", text)

Inference on Multiple GPUs

You can run MiniCPM-Llama3-V 2.5 on multiple low VRAM GPUs (12 GB or 16 GB) by distributing the model’s layers across multiple GPUs. Please refer to this tutorial for detailed instructions on how to load the model and inference using multiple low VRAM GPUs.

Inference on Mac

Click to view an example, to run MiniCPM-Llama3-V 2.5 on 💻 Mac with MPS (Apple silicon or AMD GPUs).

# test.py  Need more than 16GB memory.
import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained('openbmb/MiniCPM-Llama3-V-2_5', trust_remote_code=True, low_cpu_mem_usage=True)
model = model.to(device='mps')

tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-Llama3-V-2_5', trust_remote_code=True)
model.eval()

image = Image.open('./assets/hk_OCR.jpg').convert('RGB')
question = 'Where is this photo taken?'
msgs = [{'role': 'user', 'content': question}]

answer, context, _ = model.chat(
    image=image,
    msgs=msgs,
    context=None,
    tokenizer=tokenizer,
    sampling=True
)
print(answer)

Run with command:

`1`	`PYTORCH_ENABLE_MPS_FALLBACK=1 python test.py`

Efficient Inference with llama.cpp, Ollama, vLLM

See our fork of llama.cpp for more detail. This implementation supports smooth inference of 16~18 token/s on iPad (test environment：iPad Pro + M4).

See our fork of Ollama for more detail. This implementation supports smooth inference of 16~18 token/s on iPad (test environment：iPad Pro + M4).

vLLM now officially supports MiniCPM-V 2.6, MiniCPM-Llama3-V 2.5 and MiniCPM-V 2.0. And you can use our fork to run MiniCPM-o 2.6 for now. Click to see.

Install vLLM(>=0.7.1):

`1`	`pip install vllm`

Run Example:

Fine-tuning

Simple Fine-tuning

We support simple fine-tuning with Hugging Face for MiniCPM-o 2.6, MiniCPM-V 2.6, MiniCPM-Llama3-V 2.5 and MiniCPM-V 2.0.

Reference Document

With Align-Anything

We support fine-tuning MiniCPM-o 2.6 by PKU-Alignment Team (both vision and audio, SFT and DPO) with the Align-Anything framework. Align-Anything is a scalable framework that aims to align any-modality large models with human intentions, open-sourcing the datasets, models and benchmarks. Benefiting from its concise and modular design, it supports 30+ open-source benchmarks, 40+ models and algorithms including SFT, SimPO, RLHF, etc. It also provides 30+ directly runnable scripts, making it suitable for beginners to quickly get started.

Best Practices: MiniCPM-o 2.6.

With LLaMA-Factory

We support fine-tuning MiniCPM-o 2.6 and MiniCPM-V 2.6 with the LLaMA-Factory framework. LLaMA-Factory provides a solution for flexibly customizing the fine-tuning (Lora/Full/Qlora) of 200+ LLMs without the need for coding through the built-in web UI LLaMABoard. It supports various training methods like sft/ppo/dpo/kto and advanced algorithms like Galore/BAdam/LLaMA-Pro/Pissa/LongLoRA.

Best Practices: MiniCPM-o 2.6 | MiniCPM-V 2.6.

With the SWIFT Framework

We now support MiniCPM-V series fine-tuning with the SWIFT framework. SWIFT supports training, inference, evaluation and deployment of nearly 200 LLMs and MLLMs . It supports the lightweight training solutions provided by PEFT and a complete Adapters Library including techniques such as NEFTune, LoRA+ and LLaMA-PRO.

Best Practices：MiniCPM-V 1.0, MiniCPM-V 2.0, MiniCPM-V 2.6.

Awesome work using MiniCPM-V & MiniCPM-o

text-extract-api: Document extraction API using OCRs and Ollama supported models
comfyui_LLM_party: Build LLM workflows and integrate into existing image workflows
Ollama-OCR: OCR package uses vlms through Ollama to extract text from images and PDF
comfyui-mixlab-nodes: ComfyUI node suite supports Workflow-to-APP、GPT&3D and more
OpenAvatarChat: Interactive digital human conversation implementation on single PC
pensieve: A privacy-focused passive recording project by recording screen content
paperless-gpt: Use LLMs to handle paperless-ngx, AI-powered titles, tags and OCR
Neuro: A recreation of Neuro-Sama, but running on local models on consumer hardware

FAQs

Click here to view the FAQs

Limitations

As an experimental trial, we find MiniCPM-o 2.6 has notable limitations worth further investigation and improvement.

Unstable speech output. The speech generation can be flawed with noisy backgrounds and unmeaningful sounds.
Repeated response. The model tends to repeat its response when encountering similar consecutive user queries.
High-latency on Web Demo. Users may experience unusual high-latency when using web demo hosted on overseas servers. We recommend deploying the demo locally or with good network connections.

Model License

This repository is released under the Apache-2.0 License.
The usage of MiniCPM-o/V model weights must strictly follow MiniCPM Model License.md.
The models and weights of MiniCPM are completely free for academic research. after filling out a “questionnaire” for registration, are also available for free commercial use.

Statement

As MLLMs, MiniCPM-o/V models generate content by learning a large number of multimodal corpora, but they cannot comprehend, express personal opinions, or make value judgements. Anything generated by MiniCPM-o/V models does not represent the views and positions of the model developers

We will not be liable for any problems arising from the use of MiniCPM-o/V models, including but not limited to data security issues, risk of public opinion, or any risks and problems arising from the misdirection, misuse, dissemination, or misuse of the model.

Institutions

This project is developed by the following institutions:

🌟 Star History

Key Techniques and Other Multimodal Projects

👏 Welcome to explore key techniques of MiniCPM-o/V and other multimodal projects of our team:

VisCPM | RLPR | RLHF-V | LLaVA-UHD | RLAIF-V

Citation

If you find our model/code/paper helpful, please consider citing our papers 📝 and staring us ⭐️！

@article{yao2024minicpm,
  title={MiniCPM-V: A GPT-4V Level MLLM on Your Phone},
  author={Yao, Yuan and Yu, Tianyu and Zhang, Ao and Wang, Chongyi and Cui, Junbo and Zhu, Hongji and Cai, Tianchi and Li, Haoyu and Zhao, Weilin and He, Zhihui and others},
  journal={arXiv preprint arXiv:2408.01800},
  year={2024}
}

fastapi_mcp

Sat, 16 Aug 2025 15:26:52 +0800

tadata-org/fastapi_mcp

Built by Tadata

FastAPI-MCP

Expose your FastAPI endpoints as Model Context Protocol (MCP) tools, with Auth!

Features

Authentication built in, using your existing FastAPI dependencies!
FastAPI-native: Not just another OpenAPI -> MCP converter
Zero/Minimal configuration required - just point it at your FastAPI app and it works
Preserving schemas of your request models and response models
Preserve documentation of all your endpoints, just as it is in Swagger
Flexible deployment - Mount your MCP server to the same app, or deploy separately
ASGI transport - Uses FastAPI’s ASGI interface directly for efficient communication

Hosted Solution

If you prefer a managed hosted solution check out tadata.com.

Installation

We recommend using uv, a fast Python package installer:

`1`	`uv add fastapi-mcp`

Alternatively, you can install with pip:

`1`	`pip install fastapi-mcp`

Basic Usage

The simplest way to use FastAPI-MCP is to add an MCP server directly to your FastAPI application:

from fastapi import FastAPI
from fastapi_mcp import FastApiMCP

app = FastAPI()

mcp = FastApiMCP(app)

# Mount the MCP server directly to your FastAPI app
mcp.mount()

That’s it! Your auto-generated MCP server is now available at https://app.base.url/mcp.

Documentation, Examples and Advanced Usage

FastAPI-MCP provides comprehensive documentation. Additionaly, check out the examples directory for code samples demonstrating these features in action.

FastAPI-first Approach

FastAPI-MCP is designed as a native extension of FastAPI, not just a converter that generates MCP tools from your API. This approach offers several key advantages:

Native dependencies: Secure your MCP endpoints using familiar FastAPI Depends() for authentication and authorization
ASGI transport: Communicates directly with your FastAPI app using its ASGI interface, eliminating the need for HTTP calls from the MCP to your API
Unified infrastructure: Your FastAPI app doesn’t need to run separately from the MCP server (though separate deployment is also supported)

This design philosophy ensures minimum friction when adding MCP capabilities to your existing FastAPI services.

Development and Contributing

Thank you for considering contributing to FastAPI-MCP! We encourage the community to post Issues and create Pull Requests.

Before you get started, please see our Contribution Guide.

Community

Join MCParty Slack community to connect with other MCP enthusiasts, ask questions, and share your experiences with FastAPI-MCP.

Requirements

Python 3.10+ (Recommended 3.12)
uv

License

adk-python

Sat, 09 Aug 2025 15:28:44 +0800

google/adk-python

Agent Development Kit (ADK)

An open-source, code-first Python toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.

Important Links: Docs, Samples, Java ADK & ADK Web.

Agent Development Kit (ADK) is a flexible and modular framework for developing and deploying AI agents. While optimized for Gemini and the Google ecosystem, ADK is model-agnostic, deployment-agnostic, and is built for compatibility with other frameworks. ADK was designed to make agent development feel more like software development, to make it easier for developers to create, deploy, and orchestrate agentic architectures that range from simple tasks to complex workflows.

✨ Key Features

Rich Tool Ecosystem: Utilize pre-built tools, custom functions, OpenAPI specs, or integrate existing tools to give agents diverse capabilities, all for tight integration with the Google ecosystem.
Code-First Development: Define agent logic, tools, and orchestration directly in Python for ultimate flexibility, testability, and versioning.
Modular Multi-Agent Systems: Design scalable applications by composing multiple specialized agents into flexible hierarchies.
Deploy Anywhere: Easily containerize and deploy agents on Cloud Run or scale seamlessly with Vertex AI Agent Engine.

🤖 Agent2Agent (A2A) Protocol and ADK Integration

For remote agent-to-agent communication, ADK integrates with the A2A protocol. See this example for how they can work together.

🚀 Installation

Stable Release (Recommended)

You can install the latest stable version of ADK using pip:

`1`	`pip install google-adk`

The release cadence is weekly.

This version is recommended for most users as it represents the most recent official release.

Development Version

Bug fixes and new features are merged into the main branch on GitHub first. If you need access to changes that haven’t been included in an official PyPI release yet, you can install directly from the main branch:

`1`	`pip install git+https://github.com/google/adk-python.git@main`

Note: The development version is built directly from the latest code commits. While it includes the newest fixes and features, it may also contain experimental changes or bugs not present in the stable release. Use it primarily for testing upcoming changes or accessing critical fixes before they are officially released.

📚 Documentation

Explore the full documentation for detailed guides on building, evaluating, and deploying agents:

Documentation

🏁 Feature Highlight

Define a single agent:

from google.adk.agents import Agent
from google.adk.tools import google_search

root_agent = Agent(
    name="search_assistant",
    model="gemini-2.0-flash", # Or your preferred Gemini model
    instruction="You are a helpful assistant. Answer user questions using Google Search when needed.",
    description="An assistant that can search the web.",
    tools=[google_search]
)

Define a multi-agent system:

Define a multi-agent system with coordinator agent, greeter agent, and task execution agent. Then ADK engine and the model will guide the agents works together to accomplish the task.

from google.adk.agents import LlmAgent, BaseAgent

# Define individual agents
greeter = LlmAgent(name="greeter", model="gemini-2.0-flash", ...)
task_executor = LlmAgent(name="task_executor", model="gemini-2.0-flash", ...)

# Create parent agent and assign children via sub_agents
coordinator = LlmAgent(
    name="Coordinator",
    model="gemini-2.0-flash",
    description="I coordinate greetings and tasks.",
    sub_agents=[ # Assign sub_agents here
        greeter,
        task_executor
    ]
)

Development UI

A built-in development UI to help you test, evaluate, debug, and showcase your agent(s).

Evaluate Agents

1
2
3

adk eval \
    samples_for_testing/hello_world \
    samples_for_testing/hello_world/hello_world_eval_set_001.evalset.json

🤝 Contributing

We welcome contributions from the community! Whether it’s bug reports, feature requests, documentation improvements, or code contributions, please see our

General contribution guideline and flow.
Then if you want to contribute code, please read Code Contributing Guidelines to get started.

Vibe Coding

If you are to develop agent via vibe coding the llms.txt and the llms-full.txt can be used as context to LLM. While the former one is a summarized one and the later one has the full information in case your LLM has big enough context window.

📄 License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

Happy Agent Building!

reflex

Wed, 06 Aug 2025 15:36:08 +0800

reflex-dev/reflex

✨ Performant, customizable web apps in pure Python. Deploy in seconds. ✨

[!NOTE] 🚀 Try Reflex Build – our AI-powered app builder that generates full-stack Reflex applications in seconds.

Introduction

Reflex is a library to build full-stack web apps in pure Python.

Key features:

Pure Python - Write your app’s frontend and backend all in Python, no need to learn Javascript.
Full Flexibility - Reflex is easy to get started with, but can also scale to complex apps.
Deploy Instantly - After building, deploy your app with a single command or host it on your own server.

See our architecture page to learn how Reflex works under the hood.

⚙️ Installation

Open a terminal and run (Requires Python 3.10+):

`1`	`pip install reflex`

🥳 Create your first app

Installing reflex also installs the reflex command line tool.

Test that the install was successful by creating a new project. (Replace my_app_name with your project name):

1
2
3

mkdir my_app_name
cd my_app_name
reflex init

This command initializes a template app in your new directory.

You can run this app in development mode:

`1`	`reflex run`

You should see your app running at http://localhost:3000.

Now you can modify the source code in my_app_name/my_app_name.py. Reflex has fast refreshes so you can see your changes instantly when you save your code.

🫧 Example App

Let’s go over an example: creating an image generation UI around DALL·E. For simplicity, we just call the OpenAI API, but you could replace this with an ML model run locally.

Here is the complete code to create this. This is all done in one Python file!

import reflex as rx
import openai

openai_client = openai.OpenAI()


class State(rx.State):
    """The app state."""

    prompt = ""
    image_url = ""
    processing = False
    complete = False

    def get_image(self):
        """Get the image from the prompt."""
        if self.prompt == "":
            return rx.window_alert("Prompt Empty")

        self.processing, self.complete = True, False
        yield
        response = openai_client.images.generate(
            prompt=self.prompt, n=1, size="1024x1024"
        )
        self.image_url = response.data[0].url
        self.processing, self.complete = False, True


def index():
    return rx.center(
        rx.vstack(
            rx.heading("DALL-E", font_size="1.5em"),
            rx.input(
                placeholder="Enter a prompt..",
                on_blur=State.set_prompt,
                width="25em",
            ),
            rx.button(
                "Generate Image",
                on_click=State.get_image,
                width="25em",
                loading=State.processing
            ),
            rx.cond(
                State.complete,
                rx.image(src=State.image_url, width="20em"),
            ),
            align="center",
        ),
        width="100%",
        height="100vh",
    )

# Add state and page to the app.
app = rx.App()
app.add_page(index, title="Reflex:DALL-E")

Let’s break this down.

Reflex UI

Let’s start with the UI.

def index():
    return rx.center(
        ...
    )

This index function defines the frontend of the app.

We use different components such as center, vstack, input, and button to build the frontend. Components can be nested within each other to create complex layouts. And you can use keyword args to style them with the full power of CSS.

Reflex comes with 60+ built-in components to help you get started. We are actively adding more components, and it’s easy to create your own components.

State

Reflex represents your UI as a function of your state.

class State(rx.State):
    """The app state."""
    prompt = ""
    image_url = ""
    processing = False
    complete = False

The state defines all the variables (called vars) in an app that can change and the functions that change them.

Here the state is comprised of a prompt and image_url. There are also the booleans processing and complete to indicate when to disable the button (during image generation) and when to show the resulting image.

Event Handlers

def get_image(self):
    """Get the image from the prompt."""
    if self.prompt == "":
        return rx.window_alert("Prompt Empty")

    self.processing, self.complete = True, False
    yield
    response = openai_client.images.generate(
        prompt=self.prompt, n=1, size="1024x1024"
    )
    self.image_url = response.data[0].url
    self.processing, self.complete = False, True

Within the state, we define functions called event handlers that change the state vars. Event handlers are the way that we can modify the state in Reflex. They can be called in response to user actions, such as clicking a button or typing in a text box. These actions are called events.

Our DALL·E app has an event handler, get_image which gets this image from the OpenAI API. Using yield in the middle of an event handler will cause the UI to update. Otherwise the UI will update at the end of the event handler.

Routing

Finally, we define our app.

`1`	`app = rx.App()`

We add a page from the root of the app to the index component. We also add a title that will show up in the page preview/browser tab.

`1`	`app.add_page(index, title="DALL-E")`

You can create a multi-page app by adding more pages.

📑 Resources

📑 Docs | 🗞️ Blog | 📱 Component Library | 🖼️ Templates | 🛸 Deployment

✅ Status

Reflex launched in December 2022 with the name Pynecone.

🚀 Introducing Reflex Build — Our AI-Powered Builder Reflex Build uses AI to generate complete full-stack Python applications. It helps you quickly create, customize, and refine your Reflex apps — from frontend components to backend logic — so you can focus on your ideas instead of boilerplate code. Whether you’re prototyping or scaling, Reflex Build accelerates development by intelligently scaffolding and optimizing your app’s entire stack.

Alongside this, Reflex Cloud launched in 2025 to offer the best hosting experience for your Reflex apps. We’re continuously improving the platform with new features and capabilities.

Reflex has new releases and features coming every week! Make sure to :star: star and :eyes: watch this repository to stay up to date.

Contributing

We welcome contributions of any size! Below are some good ways to get started in the Reflex community.

Join Our Discord: Our Discord is the best place to get help on your Reflex project and to discuss how you can contribute.
GitHub Discussions: A great way to talk about features you want added or things that are confusing/need clarification.
GitHub Issues: Issues are an excellent way to report bugs. Additionally, you can try and solve an existing issue and submit a PR.

We are actively looking for contributors, no matter your skill level or experience. To contribute check out CONTRIBUTING.md

All Thanks To Our Contributors:

License

Reflex is open-source and licensed under the Apache License 2.0.

awesome-claude-code

Wed, 23 Jul 2025 15:36:28 +0800

hesreallyhim/awesome-claude-code

 █████┐ ██┐    ██┐███████┐███████┐ ██████┐ ███┐   ███┐███████┐
██┌──██┐██│    ██│██┌────┘██┌────┘██┌───██┐████┐ ████│██┌────┘
███████│██│ █┐ ██│█████┐  ███████┐██│   ██│██┌████┌██│█████┐
██┌──██│██│███┐██│██┌──┘  └────██│██│   ██│██│└██┌┘██│██┌──┘
██│  ██│└███┌███┌┘███████┐███████│└██████┌┘██│ └─┘ ██│███████┐
└─┘  └─┘ └──┘└──┘ └──────┘└──────┘ └─────┘ └─┘     └─┘└──────┘

 ────────────────────────────────────────────────────────────────────────────────────

 ██████┐██┐      █████┐ ██┐   ██┐██████┐ ███████┐     ██████┐ ██████┐ ██████┐ ███████┐
██┌────┘██│     ██┌──██┐██│   ██│██┌──██┐██┌────┘    ██┌────┘██┌───██┐██┌──██┐██┌────┘
██│     ██│     ███████│██│   ██│██│  ██│█████┐      ██│     ██│   ██│██│  ██│█████┐
██│     ██│     ██┌──██│██│   ██│██│  ██│██┌──┘      ██│     ██│   ██│██│  ██│██┌──┘
└██████┐███████┐██│  ██│└██████┌┘██████┌┘███████┐    └██████┐└██████┌┘██████┌┘███████┐
 └─────┘└──────┘└─┘  └─┘ └─────┘ └─────┘ └──────┘     └─────┘ └─────┘ └─────┘ └──────┘

This is a curated list of slash-commands, CLAUDE.md files, CLI tools, and other resources and guides for enhancing your Claude Code workflow, productivity, and vibes.

Claude Code is a cutting-edge CLI-based coding assistant and agent that you can access in your terminal or IDE. It is a rapidly evolving tool that offers a number of powerful capabilities, and allows for a lot of configuration, in a lot of different ways. Users are actively working out best practices and workflows. It is the hope that this repo will help the community share knowledge and understand how to get the most out of Claude Code.

Announcements

2025-07-18 - I ended up over-engineering the submission workflow, but I think it’s done, I just have to smoke test it and update the docs. For anyone with existing PR’s, don’t worry about updating them (for formatting purposes, that is), I can take care of it myself. For anoyne with new PR’s, you should be able to run make submit from the root directory of your fork for an interactive experience (as I said, needs smoke testing) - alternatively, add your entry to the bottom of THE_RESOURCES_TABLE and run make generate to automatically update the README.md based on the information you filled in. If it’s not working, just open a PR with the relevant information and I’ll deal with it, I created this mess anyway 😃.

▪     Workflows & Knowledge Guides
▪     Tooling
          ▫     IDE Integrations
▪     Hooks
▪     Slash-Commands
          ▫     Version Control & Git
          ▫     Code Analysis & Testing
          ▫     Context Loading & Priming
          ▫     Documentation & Changelogs
          ▫     CI / Deployment
          ▫     Project & Task Management
          ▫     Miscellaneous
▪     CLAUDE.md Files
          ▫     Language-Specific
          ▫     Domain-Specific
          ▫     Project Scaffolding & MCP
▪     Official Documentation

Workflows & Knowledge Guides 🧠

A workflow is a tightly coupled set of Claude Code-native resources that facilitate specific projects

Blogging Platform Instructions by cloudartisan ⚖️ CC-BY-SA-4.0
Provides a well-structured set of commands for publishing and maintaining a blogging platform, including commands for creating posts, managing categories, and handling media files.

ClaudeLog by InventorBlack
A comprehensive knowledge repository that features detailed breakdowns of advanced Claude Code mechanics including CLAUDE.md best practices, practical technique guides like plan mode, and a configuration guide.

Context Priming by disler
Provides a systematic approach to priming Claude Code with comprehensive project context through specialized commands for different project scenarios and development contexts.

n8n_agent by kingler
Amazing comprehensive set of comments for code analysis, QA, design, documentation, project structure, project management, optimization, and many more.

Project Bootstrapping and Task Management by steadycursor
Provides a structured set of commands for bootstrapping and managing a new project, including meta-commands for creating and editing custom slash-commands.

Project Management, Implementation, Planning, and Release by scopecraft
Really comprehensive set of commands for all aspects of SDLC.

Project Workflow System by harperreed
A set of commands that provide a comprehensive workflow system for managing projects, including task management, code review, and deployment processes.

Shipping Real Code w/ Claude by Diwank
A detailed blog post explaining the author’s process for shipping a product with Claude Code, including CLAUDE.md files and other interesting resources.

Simone by Helmi ⚖️ MIT
A broader project management workflow for Claude Code that encompasses not just a set of commands, but a system of documents, guidelines, and processes to facilitate project planning and execution.

Slash-commands megalist by wcygan
A pretty stunning list (88 at the time of this post!) of slash-commands ranging from agent orchestration, code review, project management, security, documentation, self-assessment, almost anything you can dream of.

Tooling 🧰

Tooling denotes applications that are built on top of Claude Code and consist of more components than slash-commands and CLAUDE.md files

CC Usage by ryoppippi ⚖️ MIT
Handy CLI tool for managing and analyzing Claude Code usage, based on analyzing local Claude Code logs. Presents a nice dashboard regarding cost information, token consumption, etc.

ccexp by nyatinte ⚖️ MIT
Interactive CLI tool for discovering and managing Claude Code configuration files and slash commands with a beautiful terminal UI.

Claude Code Flow by ruvnet ⚖️ MIT
This mode serves as a code-first orchestration layer, enabling Claude to write, edit, test, and optimize code autonomously across recursive agent cycles.

Claude Composer by Mike Bannister ⚖️ Unlicense
A tool that adds small enhancements to Claude Code.

Claude Hub by Claude Did This
A webhook service that connects Claude Code to GitHub repositories, enabling AI-powered code assistance directly through pull requests and issues. This integration allows Claude to analyze repositories, answer technical questions, and help developers understand and improve their codebase through simple @mentions.

Claude Squad by smtg-ai ⚖️ AGPL-3.0
Claude Squad is a terminal app that manages multiple Claude Code, Codex (and other local agents including Aider) in separate workspaces, allowing you to work on multiple tasks simultaneously.

Claude Swarm by parruda ⚖️ MIT
Launch Claude Code session that is connected to a swarm of Claude Code Agents.

Claude Task Master by eyaltoledano ⚖️ NOASSERTION
A task management system for AI-driven development with Claude, designed to work seamlessly with Cursor AI.

Claude Task Runner by grahama1970
A specialized tool to manage context isolation and focused task execution with Claude Code, solving the critical challenge of context length limitations and task focus when working with Claude on complex, multi-step projects.

Container Use by dagger ⚖️ Apache-2.0
Development environments for coding agents. Enable multiple agents to work safely and independently with your preferred stack.

IDE Integrations

claude-code.el by stevemolitor ⚖️ Apache-2.0
An Emacs interface for Claude Code CLI.

claude-code.nvim by greggh ⚖️ MIT
A seamless integration between Claude Code AI assistant and Neovim.

crystal by stravu ⚖️ MIT
A full-fledged desktop application for orchestrating, monitoring, and interacting with Claude Code agents.

Hooks 🪝

Hooks are a brand new API for Claude Code that allows users to activate commands and run scripts at different points in Claude’s agentic lifecycle.

[Experimental] - The resources listed in this section have not been fully vetted and may not work as expected, given the bleeding-edge nature of Claude Code hooks. Nevertheless, I wished to include them at least as a source of inspiration and to explore this unknown terrain. YMMV!

claude-code-hooks-sdk by beyondcode ⚖️ MIT
A Laravel-inspired PHP SDK for building Claude Code hook responses with a clean, fluent API. This SDK makes it easy to create structured JSON responses for Claude Code hooks using an expressive, chainable interface.

claude-hooks by John Lindquist ⚖️ MIT
A TypeScript-based system for configuring and customizing Claude Code hooks with a powerful and flexible interface.

Linting, testing, and notifications (in go) by Josh Symonds ⚖️ MIT
Nice set of hooks for enforcing code quality (linting, testing, notifications), with a nice configuration setup as well.

TDD Guard by Nizar Selander ⚖️ MIT
A hooks-driven system that monitors file operations in real-time and blocks changes that violate TDD principles.

Slash-Commands 🔪

Version Control & Git

/bug-fix by danielscholl
Streamlines bug fixing by creating a GitHub issue first, then a feature branch for implementing and thoroughly testing the solution before merging.

/commit by evmts ⚖️ MIT
Creates git commits using conventional commit format with appropriate emojis, following project standards and creating descriptive messages that explain the purpose of changes.

/commit-fast by steadycursor
Automates git commit process by selecting the first suggested message, generating structured commits with consistent formatting while skipping manual confirmation and removing Claude co-Contributorship footer

/create-pr by toyamarinyon ⚖️ Apache-2.0
Streamlines pull request creation by handling the entire workflow: creating a new branch, committing changes, formatting modified files with Biome, and submitting the PR.

/create-pull-request by liam-hq ⚖️ Apache-2.0
Provides comprehensive PR creation guidance with GitHub CLI, enforcing title conventions, following template structure, and offering concrete command examples with best practices.

/create-worktrees by evmts ⚖️ MIT
Creates git worktrees for all open PRs or specific branches, handling branches with slashes, cleaning up stale worktrees, and supporting custom branch creation for development.

/fix-github-issue by jeremymailen ⚖️ Apache-2.0
Analyzes and fixes GitHub issues using a structured approach with GitHub CLI for issue details, implementing necessary code changes, running tests, and creating proper commit messages.

/fix-issue by metabase ⚖️ NOASSERTION
Addresses GitHub issues by taking issue number as parameter, analyzing context, implementing solution, and testing/validating the fix for proper integration.

/fix-pr by metabase ⚖️ NOASSERTION
Fetches and fixes unresolved PR comments by automatically retrieving feedback, addressing reviewer concerns, making targeted code improvements, and streamlining the review process.

/husky by evmts ⚖️ MIT
Sets up and manages Husky Git hooks by configuring pre-commit hooks, establishing commit message standards, integrating with linting tools, and ensuring code quality on commits.

/pr-review by arkavo-org ⚖️ MIT
Reviews pull request changes to provide feedback, check for issues, and suggest improvements before merging into the main codebase.

/update-branch-name by giselles-ai ⚖️ Apache-2.0
Updates branch names with proper prefixes and formats, enforcing naming conventions, supporting semantic prefixes, and managing remote branch updates.

Code Analysis & Testing

/check by rygwdn
Performs comprehensive code quality and security checks, featuring static analysis integration, security vulnerability scanning, code style enforcement, and detailed reporting.

/clean by Graphlet-AI ⚖️ Apache-2.0
Addresses code formatting and quality issues by fixing black formatting problems, organizing imports with isort, resolving flake8 linting issues, and correcting mypy type errors.

/code_analysis by kingler
Provides a menu of advanced code analysis commands for deep inspection, including knowledge graph generation, optimization suggestions, and quality evaluation.

/optimize by to4iki ⚖️ MIT
Analyzes code performance to identify bottlenecks, proposing concrete optimizations with implementation guidance for improved application performance.

/repro-issue by rzykov ⚖️ NOASSERTION
Creates reproducible test cases for GitHub issues, ensuring tests fail reliably and documenting clear reproduction steps for developers.

/tdd by zscott
Guides development using Test-Driven Development principles, enforcing Red-Green-Refactor discipline, integrating with git workflow, and managing PR creation.

Context Loading & Priming

/context-prime by elizaOS ⚖️ MIT
Primes Claude with comprehensive project understanding by loading repository structure, setting development context, establishing project goals, and defining collaboration parameters.

/initref by okuvshynov ⚖️ MIT
Initializes reference documentation structure with standard doc templates, API reference setup, documentation conventions, and placeholder content generation.

/load-llms-txt by ethpandaops ⚖️ MIT
Loads LLM configuration files to context, importing specific terminology, model configurations, and establishing baseline terminology for AI discussions.

/load_coo_context by Mjvolk3
References specific files for sparse matrix operations, explains transform usage, compares with previous approaches, and sets data formatting context for development.

/load_dango_pipeline by Mjvolk3
Sets context for model training by referencing pipeline files, establishing working context, and preparing for pipeline work with relevant documentation.

/prime by yzyydev
Sets up initial project context by viewing directory structure and reading key files, creating standardized context with directory visualization and key documentation focus.

/rsi by ddisisto
Reads all commands and key project files to optimize AI-assisted development by streamlining the process, loading command context, and setting up for better development workflow.

Documentation & Changelogs

/add-to-changelog by berrydev-ai ⚖️ MIT
Adds new entries to changelog files while maintaining format consistency, properly documenting changes, and following established project standards for version tracking.

/create-docs by jerseycheese ⚖️ MIT
Analyzes code structure and purpose to create comprehensive documentation detailing inputs/outputs, behavior, user interaction flows, and edge cases with error handling.

/docs by slunsford
Generates comprehensive documentation that follows project structure, documenting APIs and usage patterns with consistent formatting for better user understanding.

/explain-issue-fix by hackdays-io
Documents solution approaches for GitHub issues, explaining technical decisions, detailing challenges overcome, and providing implementation context for better understanding.

/update-docs by Consiliency ⚖️ MIT
Reviews current documentation status, updates implementation progress, reviews phase documents, and maintains documentation consistency across the project.

CI / Deployment

/release by kelp ⚖️ MIT
Manages software releases by updating changelogs, reviewing README changes, evaluating version increments, and documenting release changes for better version tracking.

/run-ci by hackdays-io
Activates virtual environments, runs CI-compatible check scripts, iteratively fixes errors, and ensures all tests pass before completion.

Project & Task Management

/create-command by scopecraft
Guides Claude through creating new custom commands with proper structure by analyzing requirements, templating commands by category, enforcing command standards, and creating supporting documentation.

/create-jtbd by taddyorg ⚖️ AGPL-3.0
Creates Jobs-to-be-Done frameworks that outline user needs with structured format, focusing on specific user problems and organizing by job categories for product development.

/create-prd by taddyorg ⚖️ AGPL-3.0
Generates comprehensive product requirement documents outlining detailed specifications, requirements, and features following standardized document structure and format.

/create-prp by Wirasm ⚖️ MIT
Creates product requirement plans by reading PRP methodology, following template structure, creating comprehensive requirements, and structuring product definitions for development.

/project_hello_w_name by disler
Creates customizable greeting components with name input, demonstrating argument passing, component reusability, state management, and user input handling.

/todo by chrisleyva ⚖️ MIT
A convenient command to quickly manage project todo items without leaving the Claude Code interface, featuring due dates, sorting, task prioritization, and comprehensive todo list management.

Miscellaneous

/five by TuckerTucker
Applies the “five whys” methodology to perform root cause analysis, identify underlying issues, and create solution approaches for complex problems.

/fixing_go_in_graph by Mjvolk3
Focuses on Gene Ontology annotation integration in graph databases, handling multiple data sources, addressing graph representation issues, and ensuring correct data incorporation.

/mermaid by GaloyMoney ⚖️ NOASSERTION
Generates Mermaid diagrams from SQL schema files, creating entity relationship diagrams with table properties, validating diagram compilation, and ensuring complete entity coverage.

/review_dcell_model by Mjvolk3
Reviews old Dcell implementation files, comparing with newer Dango model, noting changes over time, and analyzing refactoring approaches for better code organization.

/use-stepper by zuplo
Reformats documentation to use React Stepper component, transforming heading formats, applying proper indentation, and maintaining markdown compatibility with admonition formatting.

CLAUDE.md Files 📂

CLAUDE.md files are files that contain important guidelines and context-specfic information or instructions that help Claude Code to better understand your project and your coding standards

Language-Specific

AI IntelliJ Plugin by didalgolab ⚖️ Apache-2.0
Provides comprehensive Gradle commands for IntelliJ plugin development with platform-specific coding patterns, detailed package structure guidelines, and clear internationalization standards.

AWS MCP Server by alexei-led ⚖️ MIT
Features multiple Python environment setup options with detailed code style guidelines, comprehensive error handling recommendations, and security considerations for AWS CLI interactions.

DroidconKotlin by touchlab ⚖️ Apache-2.0
Delivers comprehensive Gradle commands for cross-platform Kotlin Multiplatform development with clear module structure and practical guidance for dependency injection.

EDSL by expectedparrot ⚖️ MIT
Offers detailed build and test commands with strict code style enforcement, comprehensive testing requirements, and standardized development workflow using Black and mypy.

Giselle by giselles-ai ⚖️ Apache-2.0
Provides detailed build and test commands using pnpm and Vitest with strict code formatting requirements and comprehensive naming conventions for code consistency.

HASH by hashintel ⚖️ NOASSERTION
Features comprehensive repository structure breakdown with strong emphasis on coding standards, detailed Rust documentation guidelines, and systematic PR review process.

Inkline by inkline ⚖️ NOASSERTION
Structures development workflow using pnpm with emphasis on TypeScript and Vue 3 Composition API, detailed component creation process, and comprehensive testing recommendations.

JSBeeb by mattgodbolt ⚖️ GPL-3.0
Provides development guide for JavaScript BBC Micro emulator with build and testing instructions, architecture documentation, and debugging workflows.

Lamoom Python by LamoomAI ⚖️ Apache-2.0
Serves as reference for production prompt engineering library with load balancing of AI Models, API documentation, and usage patterns with examples.

LangGraphJS by langchain-ai ⚖️ MIT
Offers comprehensive build and test commands with detailed TypeScript style guidelines, layered library architecture, and monorepo structure using yarn workspaces.

Metabase by metabase ⚖️ NOASSERTION
Details workflow for REPL-driven development in Clojure/ClojureScript with emphasis on incremental development, testing, and step-by-step approach for feature implementation.

SG Cars Trends Backend by sgcarstrends
Provides comprehensive structure for TypeScript monorepo projects with detailed commands for development, testing, deployment, and AWS/Cloudflare integration.

SPy by spylang ⚖️ MIT
Enforces strict coding conventions with comprehensive testing guidelines, multiple code compilation options, and backend-specific test decorators for targeted filtering.

TPL by KarpelesLab ⚖️ MIT
Details Go project conventions with comprehensive error handling recommendations, table-driven testing approach guidelines, and modernization suggestions for latest Go features.

Domain-Specific

AVS Vibe Developer Guide by Layr-Labs ⚖️ MIT
Structures AI-assisted EigenLayer AVS development workflow with consistent naming conventions for prompt files and established terminology standards for blockchain concepts.

Comm by CommE2E ⚖️ BSD-3-Clause
Serves as a development reference for E2E-encrypted messaging applications with code organization architecture, security implementation details, and testing procedures.

Course Builder by badass-courses ⚖️ MIT
Enables real-time multiplayer capabilities for collaborative course creation with diverse tech stack integration and monorepo architecture using Turborepo.

Cursor Tools by eastlondoner ⚖️ MIT
Creates a versatile AI command interface supporting multiple providers and models with flexible command options and browser automation through “Stagehand” feature.

Guitar by soramimi ⚖️ GPL-2.0
Serves as development guide for Guitar Git GUI Client with build commands for various platforms, code style guidelines for contributing, and project structure explanation.

Network Chronicles by Fimeg ⚖️ MIT
Presents detailed implementation plan for AI-driven game characters with technical specifications for LLM integration, character guidelines, and service discovery mechanics.

Note Companion by different-ai ⚖️ MIT
Provides detailed styling isolation techniques for Obsidian plugins using Tailwind with custom prefix to prevent style conflicts and practical troubleshooting steps.

Pareto Mac by ParetoSecurity ⚖️ GPL-3.0
Serves as development guide for Mac security audit tool with build instructions, contribution guidelines, testing procedures, and workflow documentation.

SteadyStart by steadycursor
Clear and direct instructives about style, permissions, Claude’s “role”, communications, and documentation of Claude Code sessions for other team members to stay abreast.

Project Scaffolding & MCP

Basic Memory by basicmachines-co ⚖️ AGPL-3.0
Presents an innovative AI-human collaboration framework with Model Context Protocol for bidirectional LLM-markdown communication and flexible knowledge structure for complex projects.

claude-code-mcp-enhanced by grahama1970 ⚖️ MIT
Provides detailed and emphatic instructions for Claude to follow as a coding agent, with testing guidance, code examples, and compliance checks.

Perplexity MCP by Family-IT-Guy ⚖️ ISC
Offers clear step-by-step installation instructions with multiple configuration options, detailed troubleshooting guidance, and concise architecture overview of the MCP protocol.

Official Documentation 🏛️

Links to some of Anthropic’s terrific documentation and resources regarding Claude Code

Anthropic Documentation by Anthropic ⚖️ ©
The official documentation for Claude Code, including installation instructions, usage guidelines, API references, tutorials, examples, loads of information that I won’t list individually. Like Claude Code, the documentation is frequently updated.

Anthropic Quickstarts by Anthropic ⚖️ MIT
Offers comprehensive development guides for three distinct AI-powered demo projects with standardized workflows, strict code style guidelines, and containerization instructions.

Contributing 🌻

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.

Regarding content, we especially welcome:

Proven, effective resources that follow best practices and may even be in use in production.
Innovative, creative, or experimental workflows that perhaps are still being iterated upon, but have high potential value, and push the boundaries of Claude Code’s documented capabilities and use cases.
Additional libraries and tooling that are built on top of Claude Code and offer enhanced functionality.
Applications of Claude Code outside of the traditional “coding assistant” context, e.g., CI/CD integration, testing, documentation, dev-ops, etc.

See CONTRIBUTING.md for more information on how to contribute to this project. Or, fire up Claude Code and invoke the /project:add-new-resource command and let Claude walk you through it!

If you have any suggestions or thoughts on how to improve the repo, or how to best organize the list, feel free to start a Discussion topic. This is meant to be for the Claude Code community, and in general I prefer not to act on sole authority.

A note about licenses

Because simply listing a hyperlink does not qualify as redistribution, the license of the original source is not relevant to its inclusion. However, for posterity and convenience, we do host copies of all resources whose license permits it. Therefore, please include information about the resource’s license. Additionally, take note: if you do not include a LICENSE in your GitHub repo, then by default it is fully copyrighted and redistribution is not allowed. So, if you are intending to make an open source project, it’s critical to pick from one of the many available open source licenses. This is just a reminder that without a LICENSE, your project is not open source (it’s merely source-code-available) - it may of course still be included on this list, but this notice is to inform readers about the default rules regarding GitHub and LICENSE files. See here for more details.

planka

Wed, 04 Jun 2025 15:30:26 +0800

plankanban/planka

PLANKA

Project mastering driven by fun

Client demo (without server features).

⚠️ The demo GIF and client demo are based on v1 and will be updated soon.

Key Features

Collaborative Kanban Boards: Create projects, boards, lists, cards, and manage tasks with an intuitive drag-and-drop interface
Real-Time Updates: Instant syncing across all users, no refresh needed
Rich Markdown Support: Write beautifully formatted card descriptions with a powerful markdown editor
Flexible Notifications: Get alerts through 100+ providers, fully customizable to your workflow
Seamless Authentication: Single sign-on with OpenID Connect integration
Multilingual & Easy to Translate: Full internationalization support for a global audience

How to Deploy

PLANKA is easy to install using multiple methods - learn more in the installation guide.

For configuration and environment settings, see the configuration section.

Contact

Interested in a hosted version of PLANKA? Email us at github@planka.group.

For any security issues, please do not create a public issue on GitHub - instead, report it privately by emailing security@planka.group.

Note: We do NOT offer any public support via email, please use GitHub.

Join our community: Get help, share ideas, or contribute on our Discord server.

License

PLANKA is fair-code distributed under the Fair Use License and PLANKA Pro/Enterprise License.

Source Available: The source code is always visible
Self-Hostable: Deploy and host it anywhere
Extensible: Customize with your own functionality
Enterprise Licenses: Available for additional features and support

For more details, check the License Guide.

Contributing

Found a bug or have a feature request? Check out our Contributing Guide to get started.

For setting up the project locally, see the development section.

Thanks to all our contributors!

langflow

Fri, 30 May 2025 15:29:59 +0800

langflow-ai/langflow

Langflow is a powerful tool for building and deploying AI-powered agents and workflows. It provides developers with both a visual authoring experience and a built-in API server that turns every agent into an API endpoint that can be integrated into applications built on any framework or stack. Langflow comes with batteries included and supports all major LLMs, vector databases and a growing library of AI tools.

✨ Highlight features

Visual Builder to get started quickly and iterate.
Access to Code so developers can tweak any component using Python.
Playground to immediately test and iterate on their flows with step-by-step control.
Multi-agent orchestration and conversation management and retrieval.
Deploy as an API or export as JSON for Python apps.
Observability with LangSmith, LangFuse and other integrations.
Enterprise-ready security and scalability.

⚡️ Quickstart

Langflow works with Python 3.10 to 3.13.

Install with uv (recommended)

`1`	`uv pip install langflow`

Install with pip

`1`	`pip install langflow`

📦 Deployment

Self-managed

Langflow is completely open source and you can deploy it to all major deployment clouds. Follow this guide to learn how to use Docker to deploy Langflow.

Fully-managed by DataStax

DataStax Langflow is a full-managed environment with zero setup. Developers can sign up for a free account to get started.

⭐ Stay up-to-date

Star Langflow on GitHub to be instantly notified of new releases.

👋 Contribute

We welcome contributions from developers of all levels. If you’d like to contribute, please check our contributing guidelines and help make Langflow more accessible.

❤️ Contributors

hardhat

Thu, 29 May 2025 15:29:27 +0800

NomicFoundation/hardhat

packages/hardhat-core/README.md

1Panel

Mon, 21 Apr 2025 15:28:34 +0800

1Panel-dev/1Panel

Top-Rated Web-based Linux Server Management Tool

1Panel provides an intuitive web interface and MCP Server to manage websites, files, containers, databases, and LLMs on a Linux server.

Efficient Management: Through a user-friendly web graphical interface, 1Panel enables users to effortlessly manage their Linux servers. Key features include host monitoring, file management, database administration, container management, LLMs management.
Rapid Website Deployment: With deep integration of the popular open-source website building software WordPress, 1Panel streamlines the process of domain binding and SSL certificate configuration, all achievable with just one click.
Application Store: 1Panel curates a wide range of high-quality open-source tools and applications, facilitating easy installation and updates for its users.
Security and Reliability: By leveraging containerization and secure application deployment practices, 1Panel minimizes vulnerability exposure. It further enhances security through integrated firewall management and log auditing capabilities.
One-Click Backup & Restore: Data protection is made simple with 1Panel’s one-click backup and restore functionality, supporting various cloud storage solutions to ensure data integrity and availability.
MCP Server: 1Panel MCP Server allow user to execute server operations via natural language.

Quick Start

Execute the script below and follow the prompts to install 1Panel:

`1`	`curl -sSL https://resource.1panel.pro/quick_start.sh -o quick_start.sh && bash quick_start.sh`

Please refer to our documentation for more details.

中国用户请使用这个安装脚本，其应用数量比国际版本更丰富。

Screenshot

Star History

Pro Edition

Compared to the OSS Edition, 1Panel Pro Edition provides users with a wealth of enhanced features and technical support services. Enhanced features include WAF enhancement, Website monitoring, Mobile APP, custom logo and theme, etc.

Click to see Pro Edition details.

Security Information

If you discover any security issues, please refer to SECURITY.md.

License

Licensed under The GNU General Public License version 3 (GPLv3) (the “License”); you may not use this file except in compliance with the License. You may obtain a copy of the License at

https://www.gnu.org/licenses/gpl-3.0.html

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Deployment on Producthunt daily

fizzy

Fizzy

Deploying Fizzy

Fork the repo

Editing the configuration

Deploy Fizzy!

Development

Setting up

Web Push Notifications

Running tests

Database configuration

Outbound Emails

SaaS gem

Contributing

License

Product Hunt Daily | 2025-10-13

1. Meku

2. Open SaaS 2.0

3. UNDOOMED

4. Getillustrations Figma Plugin

5. AI Ultra Editing by Thumblify

6. Monitoro

7. Flow Tabs Pro

8. Minarah - Prayer Reminder App

9. AI Financial Report (10K, Q, etc)

10. Strayl

examples

Vercel Templates

For Vercelians

Adding a new example

Adding a template

The pre-commit hook

Read the Docs

Provide Feedback

MiniCPM-V

News

📌 Pinned

Contents

MiniCPM-V 4.5

Key Techniques

Evaluation

Inference Efficiency

Examples

MiniCPM-o 2.6

Evaluation

Examples

Legacy Models

MiniCPM-V & o Cookbook

Chat with Our Demo on Gradio 🤗

Online Demo

Local WebUI Demo

Inference

Model Zoo

Multi-turn Conversation

Chat with Multiple Images

In-context Few-shot Learning

Chat with Video

Speech and Audio Mode

Mimick

General Speech Conversation with Configurable Voices

Speech Conversation as an AI Assistant

Instruction-to-Speech

Voice Cloning

Addressing Various Audio Understanding Tasks

Multimodal Live Streaming

Inference on Multiple GPUs

Inference on Mac

Efficient Inference with llama.cpp, Ollama, vLLM

Fine-tuning

Simple Fine-tuning

With Align-Anything

With LLaMA-Factory

With the SWIFT Framework

Awesome work using MiniCPM-V & MiniCPM-o

FAQs

Limitations

Model License

Statement

Institutions