Today's AI Frameworks & SDKs: Fastest-Growing Projects — May 10, 2026
Today's AI Frameworks & SDKs space saw a surge in repositories focused on efficient and scalable deployment of large language models (LLMs). Many projects are exploring ways to optimize LLM performance on specific hardware, such as Apple Silicon and NVIDIA GPUs. As a result, we're seeing a growing interest in tools that enable seamless integration and customization of AI capabilities.
1bananachicken/MaaNTE, with a growth score of 71.11 and over 1,079 stars, is a notable example of this trend. This repository provides a powerful assistant tool driven by the MAAF Framework, offering users an efficient way to interact with AI models. Its rapid growth can be attributed to its versatility and ease of use, making it an attractive solution for developers seeking to integrate AI capabilities into their applications.
HKUDS-AI/polymarket-ai-trading, boasting a growth score of 61.31 and 139 stars, is another project gaining traction in the space. This repository offers an AI-assisted paper trading bot for Polymarket, utilizing OpenAI market insights and vector search capabilities to inform trading decisions. Its growth can be attributed to its innovative application of AI in finance, providing users with a unique tool for navigating complex markets.
noonghunna/club-3090, with a growth score of 59.92 and over 703 stars, is a community-driven repository focused on optimizing LLM performance on NVIDIA RTX 3090 hardware. By providing multi-engine support and model-agnostic configurations, this project has become a go-to resource for developers seeking to squeeze the most out of their hardware. Its growth can be attributed to its collaborative nature and the growing demand for efficient LLM deployment.
warpdot-dev/craft-agents-oss, sporting a growth score of 56.06 and 217 stars, is an open-source SDK for crafting AI-powered agents. This repository provides a comprehensive suite of tools for building desktop AI applications, including support for multiple LLMs and OAuth authentication. Its growth can be attributed to its flexibility and the growing interest in developing custom AI-powered solutions.
alash3al/stash, with a growth score of 42.03 and over 676 stars, offers a persistent memory layer for AI agents, enabling efficient storage and retrieval of episodes, facts, and working context. This repository's growth can be attributed to its innovative approach to AI data management and the growing need for scalable AI solutions.
iFurySt/open-codex-computer-use, boasting a growth score of 33.43 and over 698 stars, provides an open-source alternative to Codex Computer Use. By offering a flexible and customizable solution, this repository has become a popular choice among developers seeking to integrate AI capabilities into their applications. Its growth can be attributed to its accessibility and the growing demand for open-source AI solutions.
oritera/Cairn, with a growth score of 32.07 and over 804 stars, is a general-purpose state-space search engine that has been validated on autonomous penetration testing tasks. This repository's growth can be attributed to its innovative approach to search engine development and the growing interest in applying AI to complex problem-solving.
TheTom/vllm-swift, sporting a growth score of 26.56 and over 252 stars, offers high-performance LLM inference on Apple Silicon hardware using Metal plugins powered by mlx-swift. This repository's growth can be attributed to its innovative approach to optimizing LLM performance on specific hardware and the growing demand for efficient AI solutions.
Mininglamp-AI/cider, with a growth score of 21.92 and over 318 stars, provides an optimized solution for W8A8/W4A8 inference on Apple Silicon hardware, enabling faster LLM prefill and improved overall performance. This repository's growth can be attributed to its innovative approach to optimizing AI performance on specific hardware and the growing demand for efficient AI solutions.
Note that ZhiYi-R/moon-bridge has been skipped due to its unclear description.
1bananachicken/MaaNTE, with a growth score of 71.11 and over 1,079 stars, is a notable example of this trend. This repository provides a powerful assistant tool driven by the MAAF Framework, offering users an efficient way to interact with AI models. Its rapid growth can be attributed to its versatility and ease of use, making it an attractive solution for developers seeking to integrate AI capabilities into their applications.
HKUDS-AI/polymarket-ai-trading, boasting a growth score of 61.31 and 139 stars, is another project gaining traction in the space. This repository offers an AI-assisted paper trading bot for Polymarket, utilizing OpenAI market insights and vector search capabilities to inform trading decisions. Its growth can be attributed to its innovative application of AI in finance, providing users with a unique tool for navigating complex markets.
noonghunna/club-3090, with a growth score of 59.92 and over 703 stars, is a community-driven repository focused on optimizing LLM performance on NVIDIA RTX 3090 hardware. By providing multi-engine support and model-agnostic configurations, this project has become a go-to resource for developers seeking to squeeze the most out of their hardware. Its growth can be attributed to its collaborative nature and the growing demand for efficient LLM deployment.
warpdot-dev/craft-agents-oss, sporting a growth score of 56.06 and 217 stars, is an open-source SDK for crafting AI-powered agents. This repository provides a comprehensive suite of tools for building desktop AI applications, including support for multiple LLMs and OAuth authentication. Its growth can be attributed to its flexibility and the growing interest in developing custom AI-powered solutions.
alash3al/stash, with a growth score of 42.03 and over 676 stars, offers a persistent memory layer for AI agents, enabling efficient storage and retrieval of episodes, facts, and working context. This repository's growth can be attributed to its innovative approach to AI data management and the growing need for scalable AI solutions.
iFurySt/open-codex-computer-use, boasting a growth score of 33.43 and over 698 stars, provides an open-source alternative to Codex Computer Use. By offering a flexible and customizable solution, this repository has become a popular choice among developers seeking to integrate AI capabilities into their applications. Its growth can be attributed to its accessibility and the growing demand for open-source AI solutions.
oritera/Cairn, with a growth score of 32.07 and over 804 stars, is a general-purpose state-space search engine that has been validated on autonomous penetration testing tasks. This repository's growth can be attributed to its innovative approach to search engine development and the growing interest in applying AI to complex problem-solving.
TheTom/vllm-swift, sporting a growth score of 26.56 and over 252 stars, offers high-performance LLM inference on Apple Silicon hardware using Metal plugins powered by mlx-swift. This repository's growth can be attributed to its innovative approach to optimizing LLM performance on specific hardware and the growing demand for efficient AI solutions.
Mininglamp-AI/cider, with a growth score of 21.92 and over 318 stars, provides an optimized solution for W8A8/W4A8 inference on Apple Silicon hardware, enabling faster LLM prefill and improved overall performance. This repository's growth can be attributed to its innovative approach to optimizing AI performance on specific hardware and the growing demand for efficient AI solutions.
Note that ZhiYi-R/moon-bridge has been skipped due to its unclear description.