Today's LLM & Language Models: Fastest-Growing Projects — April 17, 2026
Today's the LLM & Language Models space, we're seeing a surge in tools and repositories focused on making large language models more accessible, efficient, and user-friendly. The trend is clear: developers are working to harness the power of LLMs while minimizing their complexity and environmental impact.
One of the fastest-growing repositories this week is arman-bd/guppylm (Growth Score: 99.63, Stars: 2,907), a ~9M parameter LLM that talks like a small fish. Its explosive growth can be attributed to its unique approach to making LLMs more engaging and conversational.
Another notable repository is hexiecs/talk-normal (Growth Score: 87.67, Stars: 1,164), which offers a system prompt that removes AI slop from any LLM, making it talk like a normal person. Its high growth score indicates a strong demand for more human-like interactions with language models.
amitshekhariitbhu/llm-internals (Growth Score: 63.60, Stars: 468) is another repository gaining traction, providing a step-by-step guide to learning LLM internals, from tokenization to attention and inference optimization. Its popularity suggests that developers are eager to dive deeper into the inner workings of language models.
sdyckjq-lab/llm-wiki-skill (Growth Score: 61.79, Stars: 805) offers a personal knowledge base construction skill based on Karpathy's llm-wiki method, supporting multiple platforms. Its growth can be attributed to its ability to provide users with a comprehensive and organized way of managing their knowledge.
mnfst/awesome-free-llm-apis (Growth Score: 53.67, Stars: 2,196) has been a staple in the LLM community, providing a list of permanent free LLM APIs. Its enduring popularity highlights the ongoing demand for accessible language model resources.
kessler/gemma-gem (Growth Score: 45.42, Stars: 727) allows users to run Google's Gemma 4 model entirely on-device via WebGPU, eliminating the need for API keys or cloud services. Its growth indicates a growing interest in on-device LLM processing.
askalf/dario (Growth Score: 41.56, Stars: 70) is a local LLM router that supports multiple providers and endpoints, making it an attractive solution for developers looking to simplify their workflow. Despite its relatively low star count, its high growth score suggests significant potential.
Pratiyush/llm-wiki (Growth Score: 41.22, Stars: 106) offers a comprehensive knowledge base constructed from Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions, following Karpathy's LLM Wiki pattern. Its growth is likely driven by its ability to provide users with a centralized hub for their knowledge.
lucasastorian/llmwiki (Growth Score: 35.54, Stars: 486) provides an open-source implementation of Karpathy's LLM Wiki, allowing users to upload documents and connect their Claude account via MCP. Its popularity can be attributed to its flexibility and ease of use.
Lastly, atomicmemory/llm-wiki-compiler (Growth Score: 34.25, Stars: 543) is a knowledge compiler that takes raw sources and produces an interlinked wiki output, inspired by Karpathy's LLM Wiki pattern. Its growth indicates a growing interest in automating knowledge management tasks using language models.
One of the fastest-growing repositories this week is arman-bd/guppylm (Growth Score: 99.63, Stars: 2,907), a ~9M parameter LLM that talks like a small fish. Its explosive growth can be attributed to its unique approach to making LLMs more engaging and conversational.
Another notable repository is hexiecs/talk-normal (Growth Score: 87.67, Stars: 1,164), which offers a system prompt that removes AI slop from any LLM, making it talk like a normal person. Its high growth score indicates a strong demand for more human-like interactions with language models.
amitshekhariitbhu/llm-internals (Growth Score: 63.60, Stars: 468) is another repository gaining traction, providing a step-by-step guide to learning LLM internals, from tokenization to attention and inference optimization. Its popularity suggests that developers are eager to dive deeper into the inner workings of language models.
sdyckjq-lab/llm-wiki-skill (Growth Score: 61.79, Stars: 805) offers a personal knowledge base construction skill based on Karpathy's llm-wiki method, supporting multiple platforms. Its growth can be attributed to its ability to provide users with a comprehensive and organized way of managing their knowledge.
mnfst/awesome-free-llm-apis (Growth Score: 53.67, Stars: 2,196) has been a staple in the LLM community, providing a list of permanent free LLM APIs. Its enduring popularity highlights the ongoing demand for accessible language model resources.
kessler/gemma-gem (Growth Score: 45.42, Stars: 727) allows users to run Google's Gemma 4 model entirely on-device via WebGPU, eliminating the need for API keys or cloud services. Its growth indicates a growing interest in on-device LLM processing.
askalf/dario (Growth Score: 41.56, Stars: 70) is a local LLM router that supports multiple providers and endpoints, making it an attractive solution for developers looking to simplify their workflow. Despite its relatively low star count, its high growth score suggests significant potential.
Pratiyush/llm-wiki (Growth Score: 41.22, Stars: 106) offers a comprehensive knowledge base constructed from Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions, following Karpathy's LLM Wiki pattern. Its growth is likely driven by its ability to provide users with a centralized hub for their knowledge.
lucasastorian/llmwiki (Growth Score: 35.54, Stars: 486) provides an open-source implementation of Karpathy's LLM Wiki, allowing users to upload documents and connect their Claude account via MCP. Its popularity can be attributed to its flexibility and ease of use.
Lastly, atomicmemory/llm-wiki-compiler (Growth Score: 34.25, Stars: 543) is a knowledge compiler that takes raw sources and produces an interlinked wiki output, inspired by Karpathy's LLM Wiki pattern. Its growth indicates a growing interest in automating knowledge management tasks using language models.