Today's LLM & Language Models: Fastest-Growing Projects — April 12, 2026
Today's the LLM & Language Models space, we're seeing a surge in innovative projects that leverage large language models to build personal knowledge bases and enhance productivity. The trend is shifting towards on-device inference, with several repositories gaining traction for their ability to run LLMs locally without relying on cloud APIs. Meanwhile, curated lists of free LLM APIs continue to attract attention from developers.
AgriciDaniel's claude-obsidian repository has taken the top spot with a growth score of 97.40 and 506 stars. This project combines Claude and Obsidian to create a persistent, compounding wiki vault based on Karpathy's LLM Wiki pattern, allowing users to save and research knowledge in a seamless manner. Its rapid growth can be attributed to its unique approach to knowledge management and the popularity of Obsidian among note-taking enthusiasts.
kessler's gemma-gem repository boasts an impressive 645 stars and a growth score of 67.71. This project runs Google's Gemma 4 model entirely on-device via WebGPU, eliminating the need for API keys or cloud services. Its growing popularity stems from its ability to provide secure, local LLM inference without compromising performance.
mnfst's awesome-free-llm-apis repository has garnered an impressive 2,104 stars and a growth score of 63.30. This curated list provides permanent free LLM API keys, making it an invaluable resource for developers looking to integrate language models into their projects. Its popularity is driven by the increasing demand for accessible LLM APIs.
sdyckjq-lab's llm-wiki-skill repository has gained significant traction with a growth score of 63.14 and 527 stars. This project builds upon Karpathy's LLM Wiki method to create a personal knowledge base skill that supports multiple platforms. Its growing popularity can be attributed to its flexibility and the widespread adoption of the LLM Wiki pattern.
Pratiyush's llm-wiki repository boasts a growth score of 60.88 and 58 stars, despite being a relatively new project. This repository implements Karpathy's LLM Wiki pattern to create an LLM-powered knowledge base from various sources, including Claude Code and Copilot sessions. Its rapid growth is driven by the increasing interest in personal knowledge management using LLMs.
Other notable projects this week include xoai's sage-wiki, which compiles personal knowledge bases into structured, interlinked wikis; quantumaikr's quant.cpp, which enables lossless KV cache compression and single-header libraries for LLM inference; and m0at's rvllm, which provides high-performance LLM inference in Rust. These projects demonstrate the diverse range of innovations happening in the LLM & Language Models space.
Lastly, rasbt's llm-architecture-gallery repository remains a valuable resource with 1,020 stars, despite its relatively low growth score. This project provides source data for LLM architecture galleries, serving as a comprehensive reference point for developers and researchers alike.
AgriciDaniel's claude-obsidian repository has taken the top spot with a growth score of 97.40 and 506 stars. This project combines Claude and Obsidian to create a persistent, compounding wiki vault based on Karpathy's LLM Wiki pattern, allowing users to save and research knowledge in a seamless manner. Its rapid growth can be attributed to its unique approach to knowledge management and the popularity of Obsidian among note-taking enthusiasts.
kessler's gemma-gem repository boasts an impressive 645 stars and a growth score of 67.71. This project runs Google's Gemma 4 model entirely on-device via WebGPU, eliminating the need for API keys or cloud services. Its growing popularity stems from its ability to provide secure, local LLM inference without compromising performance.
mnfst's awesome-free-llm-apis repository has garnered an impressive 2,104 stars and a growth score of 63.30. This curated list provides permanent free LLM API keys, making it an invaluable resource for developers looking to integrate language models into their projects. Its popularity is driven by the increasing demand for accessible LLM APIs.
sdyckjq-lab's llm-wiki-skill repository has gained significant traction with a growth score of 63.14 and 527 stars. This project builds upon Karpathy's LLM Wiki method to create a personal knowledge base skill that supports multiple platforms. Its growing popularity can be attributed to its flexibility and the widespread adoption of the LLM Wiki pattern.
Pratiyush's llm-wiki repository boasts a growth score of 60.88 and 58 stars, despite being a relatively new project. This repository implements Karpathy's LLM Wiki pattern to create an LLM-powered knowledge base from various sources, including Claude Code and Copilot sessions. Its rapid growth is driven by the increasing interest in personal knowledge management using LLMs.
Other notable projects this week include xoai's sage-wiki, which compiles personal knowledge bases into structured, interlinked wikis; quantumaikr's quant.cpp, which enables lossless KV cache compression and single-header libraries for LLM inference; and m0at's rvllm, which provides high-performance LLM inference in Rust. These projects demonstrate the diverse range of innovations happening in the LLM & Language Models space.
Lastly, rasbt's llm-architecture-gallery repository remains a valuable resource with 1,020 stars, despite its relatively low growth score. This project provides source data for LLM architecture galleries, serving as a comprehensive reference point for developers and researchers alike.