Today's LLM & Language Models: Fastest-Growing Projects — April 22, 2026
Today's the LLM & Language Models space, we're seeing a surge in tools that enable users to harness the power of large language models for personal knowledge management and content creation. Many of these projects are inspired by Andrew Karpathy's LLM Wiki pattern, which emphasizes the importance of compiling and structuring knowledge into a usable format. As a result, we're witnessing rapid growth in repositories that offer innovative solutions for building personal wikis, knowledge bases, and language models.
arman-bd/guppylm is a ~9M parameter LLM that talks like a small fish, with a remarkable growth score of 80.60 and over 2,981 stars. Its unique approach to generating human-like text has captured the attention of developers, making it one of the fastest-growing repositories in this category.
hexiecs/talk-normal is another standout project, boasting a growth score of 66.21 and 1,392 stars. This tool allows users to make any LLM talk like a normal person by removing AI slop, demonstrating its potential for improving language model output.
sdyckjq-lab/llm-wiki-skill has gained significant traction with a growth score of 57.71 and 990 stars. As a personal knowledge base construction skill based on Karpathy's LLM Wiki method, it supports multiple platforms and offers users a comprehensive tool for managing their knowledge.
amitshekhariitbhu/llm-internals is growing steadily with a score of 40.85 and 607 stars, providing developers with a step-by-step guide to learning LLM internals, from tokenization to attention and inference optimization.
kessler/gemma-gem has attracted attention with its innovative approach to running Google's Gemma 4 model entirely on-device via WebGPU, achieving a growth score of 35.85 and 820 stars. By eliminating the need for API keys, cloud services, or data transmission, this project offers a unique solution for developers seeking greater control over their models.
xoai/sage-wiki is an LLM-compiled personal knowledge base that has garnered significant interest with a growth score of 35.36 and 451 stars. Its ability to compile papers, articles, and notes into a structured wiki makes it an attractive tool for researchers and writers.
VectifyAI/OpenKB boasts a growth score of 32.00 and 474 stars, offering users an open LLM knowledge base that leverages the power of large language models for knowledge management.
lucasastorian/llmwiki is an open-source implementation of Karpathy's LLM Wiki pattern, growing steadily with a score of 30.92 and 606 stars. Its ability to upload documents and connect Claude accounts via MCP makes it an appealing solution for developers seeking to build their own personal wikis.
atomicmemory/llm-wiki-compiler has gained traction with its knowledge compiler, which transforms raw sources into interlinked wikis, inspired by Karpathy's LLM Wiki pattern. With a growth score of 30.26 and 681 stars, this project demonstrates the growing interest in automated knowledge management tools.
Pratiyush/llm-wiki rounds out our list with a growth score of 28.43 and 142 stars, offering users an LLM-powered knowledge base from their Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions.
arman-bd/guppylm is a ~9M parameter LLM that talks like a small fish, with a remarkable growth score of 80.60 and over 2,981 stars. Its unique approach to generating human-like text has captured the attention of developers, making it one of the fastest-growing repositories in this category.
hexiecs/talk-normal is another standout project, boasting a growth score of 66.21 and 1,392 stars. This tool allows users to make any LLM talk like a normal person by removing AI slop, demonstrating its potential for improving language model output.
sdyckjq-lab/llm-wiki-skill has gained significant traction with a growth score of 57.71 and 990 stars. As a personal knowledge base construction skill based on Karpathy's LLM Wiki method, it supports multiple platforms and offers users a comprehensive tool for managing their knowledge.
amitshekhariitbhu/llm-internals is growing steadily with a score of 40.85 and 607 stars, providing developers with a step-by-step guide to learning LLM internals, from tokenization to attention and inference optimization.
kessler/gemma-gem has attracted attention with its innovative approach to running Google's Gemma 4 model entirely on-device via WebGPU, achieving a growth score of 35.85 and 820 stars. By eliminating the need for API keys, cloud services, or data transmission, this project offers a unique solution for developers seeking greater control over their models.
xoai/sage-wiki is an LLM-compiled personal knowledge base that has garnered significant interest with a growth score of 35.36 and 451 stars. Its ability to compile papers, articles, and notes into a structured wiki makes it an attractive tool for researchers and writers.
VectifyAI/OpenKB boasts a growth score of 32.00 and 474 stars, offering users an open LLM knowledge base that leverages the power of large language models for knowledge management.
lucasastorian/llmwiki is an open-source implementation of Karpathy's LLM Wiki pattern, growing steadily with a score of 30.92 and 606 stars. Its ability to upload documents and connect Claude accounts via MCP makes it an appealing solution for developers seeking to build their own personal wikis.
atomicmemory/llm-wiki-compiler has gained traction with its knowledge compiler, which transforms raw sources into interlinked wikis, inspired by Karpathy's LLM Wiki pattern. With a growth score of 30.26 and 681 stars, this project demonstrates the growing interest in automated knowledge management tools.
Pratiyush/llm-wiki rounds out our list with a growth score of 28.43 and 142 stars, offering users an LLM-powered knowledge base from their Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions.