Today's AI Frameworks & SDKs: Fastest-Growing Projects — May 02, 2026
Today's AI Frameworks & SDKs, we're seeing a surge in interest around tools that enable efficient and scalable AI development, particularly those focused on language models, computer vision, and decentralized computing. The top-growing projects are showcasing innovative approaches to optimizing performance, reducing latency, and expanding the reach of AI capabilities.
The fastest-growing project this week is alash3al/stash, with a growth score of 79.69 and 618 stars. Stash provides a persistent memory layer for AI agents, storing episodes, facts, and working context in Postgres, making it an attractive solution for developers seeking to improve the efficiency of their AI systems. Its self-hosted, single-binary design has likely contributed to its rapid growth.
Luce-Org/lucebox-hub, with a growth score of 40.91 and 1,407 stars, is gaining traction due to its hand-tuned LLM inference capabilities optimized for specific consumer hardware. This project's focus on delivering high-performance AI inference on a range of devices has resonated with developers seeking to deploy AI models in resource-constrained environments.
Baidu-baige/LoongForge boasts a growth score of 23.94 and 152 stars, thanks to its modular, scalable, and highly efficient training framework for language, multimodal, and embodied models. As researchers and practitioners continue to push the boundaries of AI capabilities, LoongForge's flexibility and performance have made it an increasingly popular choice.
Unclebob/swarm-forge, with a growth score of 23.33 and 433 stars, offers a simple tool for coordinating several AI agents, filling a niche in the development landscape. Its ease of use and effectiveness in managing complex AI workflows have likely driven its adoption among developers.
Simoncirstoiu/alice has a growth score of 20.42 and 300 stars, thanks to its AI-powered YOLO dataset management toolkit, which provides a comprehensive solution for analyzing, learning, ingesting, curating, and exporting datasets. Alice's popularity can be attributed to the growing need for efficient data management in computer vision applications.
FonaTech/Project_Chronos boasts a growth score of 20.35 and 118 stars, featuring zero-stall MoE inference via lookahead prediction and async DMA prefetching, optimized for SSD I/O with hybrid MLA+sliding window attention. This project's innovative approach to optimizing AI performance has captured the interest of developers seeking to push the boundaries of model efficiency.
Hi-godot/godot-ai, with a growth score of 19.93 and 161 stars, provides a production-grade MCP server and AI tools for the Godot engine, demonstrating the growing demand for AI capabilities in game development and other interactive applications. Its community-driven nature has likely contributed to its adoption among developers.
416rehman/DeepZero has a growth score of 19.76 and 376 stars, thanks to its automated vulnerability research framework that parses, decompiles, and analyzes thousands of Windows kernel drivers for exploitable IOCTLs natively using AI agents. DeepZero's innovative approach to security research has made it an attractive solution for developers seeking to identify vulnerabilities.
Agent-FM/agentfm-core boasts a growth score of 18.05 and 104 stars, enabling users to run massive AI workloads directly across a global mesh of idle CPUs and GPUs. This decentralized AI supercomputer concept has resonated with developers seeking to harness the power of collective computing resources.
Lastly, jegly/Box, with a growth score of 17.77 and 161 stars, offers a private on-device AI suite for Android, featuring llama.cpp, whisper.cpp, stable-diffusion.cpp, GGUF import, voice chat, vision AI, on-device image generation, biometric lock, encrypted history, and NPU/TPU/GPU acceleration. Box's comprehensive feature set has likely driven its adoption among developers seeking to deploy AI capabilities on mobile devices.
The fastest-growing project this week is alash3al/stash, with a growth score of 79.69 and 618 stars. Stash provides a persistent memory layer for AI agents, storing episodes, facts, and working context in Postgres, making it an attractive solution for developers seeking to improve the efficiency of their AI systems. Its self-hosted, single-binary design has likely contributed to its rapid growth.
Luce-Org/lucebox-hub, with a growth score of 40.91 and 1,407 stars, is gaining traction due to its hand-tuned LLM inference capabilities optimized for specific consumer hardware. This project's focus on delivering high-performance AI inference on a range of devices has resonated with developers seeking to deploy AI models in resource-constrained environments.
Baidu-baige/LoongForge boasts a growth score of 23.94 and 152 stars, thanks to its modular, scalable, and highly efficient training framework for language, multimodal, and embodied models. As researchers and practitioners continue to push the boundaries of AI capabilities, LoongForge's flexibility and performance have made it an increasingly popular choice.
Unclebob/swarm-forge, with a growth score of 23.33 and 433 stars, offers a simple tool for coordinating several AI agents, filling a niche in the development landscape. Its ease of use and effectiveness in managing complex AI workflows have likely driven its adoption among developers.
Simoncirstoiu/alice has a growth score of 20.42 and 300 stars, thanks to its AI-powered YOLO dataset management toolkit, which provides a comprehensive solution for analyzing, learning, ingesting, curating, and exporting datasets. Alice's popularity can be attributed to the growing need for efficient data management in computer vision applications.
FonaTech/Project_Chronos boasts a growth score of 20.35 and 118 stars, featuring zero-stall MoE inference via lookahead prediction and async DMA prefetching, optimized for SSD I/O with hybrid MLA+sliding window attention. This project's innovative approach to optimizing AI performance has captured the interest of developers seeking to push the boundaries of model efficiency.
Hi-godot/godot-ai, with a growth score of 19.93 and 161 stars, provides a production-grade MCP server and AI tools for the Godot engine, demonstrating the growing demand for AI capabilities in game development and other interactive applications. Its community-driven nature has likely contributed to its adoption among developers.
416rehman/DeepZero has a growth score of 19.76 and 376 stars, thanks to its automated vulnerability research framework that parses, decompiles, and analyzes thousands of Windows kernel drivers for exploitable IOCTLs natively using AI agents. DeepZero's innovative approach to security research has made it an attractive solution for developers seeking to identify vulnerabilities.
Agent-FM/agentfm-core boasts a growth score of 18.05 and 104 stars, enabling users to run massive AI workloads directly across a global mesh of idle CPUs and GPUs. This decentralized AI supercomputer concept has resonated with developers seeking to harness the power of collective computing resources.
Lastly, jegly/Box, with a growth score of 17.77 and 161 stars, offers a private on-device AI suite for Android, featuring llama.cpp, whisper.cpp, stable-diffusion.cpp, GGUF import, voice chat, vision AI, on-device image generation, biometric lock, encrypted history, and NPU/TPU/GPU acceleration. Box's comprehensive feature set has likely driven its adoption among developers seeking to deploy AI capabilities on mobile devices.