Today's LLM & Language Models: Fastest-Growing Projects — April 13, 2026
Today's the LLM & Language Models space, we're seeing a surge in interest around tools that enable users to harness the power of large language models (LLMs) for personal knowledge management and on-device inference. The top-growing repositories are focused on making LLMs more accessible and usable, with many leveraging Karpathy's LLM Wiki pattern to create persistent, compounding wiki vaults.
AgriciDaniel's claude-obsidian repository has taken the top spot with a growth score of 94.50 and 615 stars, as it offers a Claude + Obsidian knowledge companion that allows users to create a persistent, compounding wiki vault based on Karpathy's LLM Wiki pattern. This tool is growing rapidly due to its ability to provide a seamless way for users to manage their knowledge and notes using LLMs.
Kessler's gemma-gem repository comes in second with a growth score of 62.06 and 675 stars, as it enables users to run Google's Gemma 4 model entirely on-device via WebGPU — eliminating the need for API keys, cloud services, or data leaving their machine. This tool is gaining traction due to its ability to provide fast and private LLM inference capabilities.
Mnfst's awesome-free-llm-apis repository has a growth score of 61.09 and an impressive 2,126 stars, as it provides a permanent list of free LLM API keys that users can leverage for their projects. This tool is growing rapidly due to its comprehensive collection of free resources, making it easier for developers to get started with LLMs.
Sdyckjq-lab's llm-wiki-skill repository has a growth score of 58.62 and 569 stars, as it offers a personal knowledge base construction skill based on Karpathy's llm-wiki method, supporting multiple platforms. This tool is gaining traction due to its ability to provide users with a structured way to manage their knowledge using LLMs.
Xoai's sage-wiki repository has a growth score of 57.94 and 371 stars, as it offers an LLM-compiled personal knowledge base that compiles papers, articles, and notes into a structured, interlinked wiki. This tool is growing rapidly due to its ability to provide users with a seamless way to manage their knowledge using LLMs.
Pratiyush's llm-wiki repository has a growth score of 49.80 and 63 stars, as it offers an LLM-powered knowledge base from various sources, including Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions. This tool is gaining traction due to its ability to provide users with a comprehensive way to manage their knowledge using LLMs.
Quantumaikr's quant.cpp repository has a growth score of 34.56 and 377 stars, as it offers an LLM inference engine that provides 7x longer context, pure C, zero dependencies, lossless KV cache compression + single-header library. This tool is growing rapidly due to its ability to provide fast and efficient LLM inference capabilities.
M0at's rvllm repository has a growth score of 34.06 and 420 stars, as it offers high-performance LLM inference in Rust, serving as a drop-in vLLM replacement. This tool is gaining traction due to its ability to provide users with fast and efficient LLM inference capabilities.
Zolotukhin's zinc repository has a growth score of 24.55 and 306 stars, as it offers Zig INferenCe Engine — local LLM inference on AMD GPUs and Apple Silicon. This tool is growing rapidly due to its ability to provide fast and private LLM inference capabilities.
Rasbt's llm-architecture-gallery repository rounds out the list with a growth score of 22.38 and 1,025 stars, as it offers source data for an LLM architecture gallery. While not directly related to the other tools on this list, this repository is still gaining traction due to its comprehensive collection of LLM architectures.
AgriciDaniel's claude-obsidian repository has taken the top spot with a growth score of 94.50 and 615 stars, as it offers a Claude + Obsidian knowledge companion that allows users to create a persistent, compounding wiki vault based on Karpathy's LLM Wiki pattern. This tool is growing rapidly due to its ability to provide a seamless way for users to manage their knowledge and notes using LLMs.
Kessler's gemma-gem repository comes in second with a growth score of 62.06 and 675 stars, as it enables users to run Google's Gemma 4 model entirely on-device via WebGPU — eliminating the need for API keys, cloud services, or data leaving their machine. This tool is gaining traction due to its ability to provide fast and private LLM inference capabilities.
Mnfst's awesome-free-llm-apis repository has a growth score of 61.09 and an impressive 2,126 stars, as it provides a permanent list of free LLM API keys that users can leverage for their projects. This tool is growing rapidly due to its comprehensive collection of free resources, making it easier for developers to get started with LLMs.
Sdyckjq-lab's llm-wiki-skill repository has a growth score of 58.62 and 569 stars, as it offers a personal knowledge base construction skill based on Karpathy's llm-wiki method, supporting multiple platforms. This tool is gaining traction due to its ability to provide users with a structured way to manage their knowledge using LLMs.
Xoai's sage-wiki repository has a growth score of 57.94 and 371 stars, as it offers an LLM-compiled personal knowledge base that compiles papers, articles, and notes into a structured, interlinked wiki. This tool is growing rapidly due to its ability to provide users with a seamless way to manage their knowledge using LLMs.
Pratiyush's llm-wiki repository has a growth score of 49.80 and 63 stars, as it offers an LLM-powered knowledge base from various sources, including Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions. This tool is gaining traction due to its ability to provide users with a comprehensive way to manage their knowledge using LLMs.
Quantumaikr's quant.cpp repository has a growth score of 34.56 and 377 stars, as it offers an LLM inference engine that provides 7x longer context, pure C, zero dependencies, lossless KV cache compression + single-header library. This tool is growing rapidly due to its ability to provide fast and efficient LLM inference capabilities.
M0at's rvllm repository has a growth score of 34.06 and 420 stars, as it offers high-performance LLM inference in Rust, serving as a drop-in vLLM replacement. This tool is gaining traction due to its ability to provide users with fast and efficient LLM inference capabilities.
Zolotukhin's zinc repository has a growth score of 24.55 and 306 stars, as it offers Zig INferenCe Engine — local LLM inference on AMD GPUs and Apple Silicon. This tool is growing rapidly due to its ability to provide fast and private LLM inference capabilities.
Rasbt's llm-architecture-gallery repository rounds out the list with a growth score of 22.38 and 1,025 stars, as it offers source data for an LLM architecture gallery. While not directly related to the other tools on this list, this repository is still gaining traction due to its comprehensive collection of LLM architectures.