StarBoard

Moving right now, since the previous snapshot

#RepoCategory LanguageAge StarsNow24h7d
462 Ebony-Vinyl/dsh-our-free-model ↗
在 dsh 里装上这个插件即可，无需登录、注册或填 API Key，就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型——完全免费，不限量。 All you do is install this plugin in dsh: no login, no sign-up, no API key — the frontier models are just there, DeepSeek V4.1 Flash and Kimi K3 among them. Completely free, with no usage cap.
 LLM Models, Inference & Training JavaScript 37d 54363 - +1554 +13091
1020 YuJunZhiXue/dsh-purge ↗
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开，请查看署名要求部分
 LLM Models, Inference & Training JavaScript 73d 14581 - +708 +6460
887 Niko1221/Strata ↗
Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
 LLM Models, Inference & Training C++ 37d 37090 - +579 +5077
630 tashfeenahmed/freellmapi ↗
7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
 LLM Models, Inference & Training TypeScript 193d 45200 - +515 +4224
644 shiyu-coder/Kronos ↗
Kronos: A Foundation Model for the Language of Financial Markets
 LLM Models, Inference & Training Python 487d 44683 - +512 +1100
327 Alishahryar1/free-claude-code ↗
Use Claude Code, Codex, Antigravity, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
 LLM Models, Inference & Training Python 276d 65451 - +230 +2287
872 p-e-w/heretic ↗
Fully automatic censorship removal for language models
 LLM Models, Inference & Training Python 405d 37313 - +85 +899
83 ggml-org/llama.cpp ↗
LLM inference in C/C++
 LLM Models, Inference & Training C++ 1331d 132992 - +94 +738
364 BerriAI/litellm ↗
The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]
 LLM Models, Inference & Training Python 1193d 62455 - +47 +477
712 lyogavin/airllm ↗
AirLLM 2.8T LLM inference with single 4GB GPU
 LLM Models, Inference & Training Jupyter Notebook 1237d 41434 - +86 +867
625 JustVugg/colibri ↗
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
 LLM Models, Inference & Training C 122d 45392 - +182 +1650
1229 SoMarkAI/OXR ↗
An open-source document parsing model with SOTA accuracy and speed, using an RSI loop during post-training.
 LLM Models, Inference & Training Python 26d 2761 - +91 -
339 RVC-Boss/GPT-SoVITS ↗
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
 LLM Models, Inference & Training Python 1021d 64423 - +114 +841
342 karpathy/nanoGPT ↗
The simplest, fastest repository for training/finetuning medium-sized GPTs.
 LLM Models, Inference & Training Python 1404d 64294 - +39 +237
450 microsoft/VibeVoice ↗
Open-Source Frontier Voice AI
 LLM Models, Inference & Training Python 432d 55081 - +26 +127
749 HKUDS/LightRAG ↗
[EMNLP2025] LightRAG: Simple and Fast Retrieval-Augmented Generation
 LLM Models, Inference & Training Python 759d 40454 - +10 +143
1004 deepseek-ai/DeepInfer ↗
DeepInfer: High-performance, large-scale inference framework for DeepSeek models
 LLM Models, Inference & Training Python 15d 24656 - +155 +2800
227 unslothai/unsloth ↗
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
 LLM Models, Inference & Training Python 1067d 78878 - +51 +352
527 mudler/LocalAI ↗
LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
 LLM Models, Inference & Training Go 1323d 50103 - +21 +168
733 OpenBMB/VoxCPM ↗
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
 LLM Models, Inference & Training Python 410d 40919 - +59 +399
925 Tencent/WeKnora ↗
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
 LLM Models, Inference & Training Go 466d 36123 - +60 +813
981 google-research/timesfm ↗
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
 LLM Models, Inference & Training Python 915d 34944 - +100 +182
988 NandhaKishorM/laya ↗
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
 LLM Models, Inference & Training Python 43d 34770 - +58 +565
337 jingyaogong/minimind ↗
🧠 Train a 64M-parameter LLM from scratch in just 2h!
 LLM Models, Inference & Training Python 826d 64692 - +27 +325
340 keras-team/keras ↗
Deep Learning for humans
 LLM Models, Inference & Training Python 4236d 64408 - +9 +23
401 karpathy/nanochat ↗
The best ChatGPT that $100 can buy.
 LLM Models, Inference & Training Python 383d 59179 - +49 +275
454 ggml-org/whisper.cpp ↗
Port of OpenAI's Whisper model in C/C++
 LLM Models, Inference & Training C++ 1497d 54830 - +24 +195
482 run-llama/llama_index ↗
LlamaIndex is the document processing platform for AI
 LLM Models, Inference & Training Python 1459d 52764 - +14 +120
533 GokuMohandas/Made-With-ML ↗
Learn how to develop, deploy and iterate on production-grade ML applications.
 LLM Models, Inference & Training Jupyter Notebook 2917d 49869 - +14 +59
629 janhq/jan ↗
Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.
 LLM Models, Inference & Training Rust 1171d 45204 - +21 +139
801 RVC-Project/Retrieval-based-Voice-Conversion-WebUI ↗
Easily train a good VC model with voice data <= 10 mins!
 LLM Models, Inference & Training Python 1314d 38981 - +12 +138
825 AlexsJones/llmfit ↗
Hundreds of models & providers. One command to find what runs on your hardware.
 LLM Models, Inference & Training Rust 258d 38524 - +19 +176
830 sgl-project/sglang ↗
SGLang is a high-performance serving framework for large language models and multimodal models.
 LLM Models, Inference & Training Python 1027d 38328 - +18 +525
972 SillyTavern/SillyTavern ↗
LLM Frontend for Power Users.
 LLM Models, Inference & Training JavaScript 1360d 35159 - +20 +254
1013 openai/math ↗
LLM Models, Inference & Training Lean 25d 17198 - +61 +1714
1035 jaredpalmer/kev ↗
Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on your own
 LLM Models, Inference & Training Python 44d 9737 - +16 +228
1138 nykooi1/vibe-wise ↗
A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.
 LLM Models, Inference & Training Python 32d 4089 - +16 +171
1177 ranxianglei/billion-context ↗
A context-compression plugin for month-long single sessions (billions of tokens) ,small context windows (a 100K context is enough) and token savings (5x fewer tokens) .上下文压缩插件，月级长会话会话， 兼顾小窗口(100k上下文足矣)省token(省5倍token) . billion-context is all you need
 LLM Models, Inference & Training TypeScript 86d 3511 - +65 +879
1221 mmparko/dsh-purge ↗
DeepSeek Harness 破甲 (mirror of YuJunZhiXue/dsh-purge) — 上游即将删库，此为本仓库全量镜像：含全部分支/73 个 tag/完整提交历史。原作者署名与二开要求见 README 署名章节。
 LLM Models, Inference & Training JavaScript 18d 2828 - +77 +642
1276 OpenMOSS/MOSS-Image ↗
Official repository for MOSS-Image, a 5B-parameter base DiT model with high-quality text rendering, layout generation, and efficient fine-tuning.
 LLM Models, Inference & Training Python 15d 2357 - +50 -
1281 mw00/project-maya ↗
Project Maya - run GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on your own GPU(s), from one card to 16, NVIDIA or AMD, Linux or Windows. Engine, server, dashboard, OpenAI- and Anthropic-compatible API, quant tools.
 LLM Models, Inference & Training Cuda 25d 2331 - +48 -

LLM Models, Inference & Training, movers in the last 24 hours

#RepoCategory LanguageAge StarsNow24h7d
462 Ebony-Vinyl/dsh-our-free-model ↗
在 dsh 里装上这个插件即可，无需登录、注册或填 API Key，就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型——完全免费，不限量。 All you do is install this plugin in dsh: no login, no sign-up, no API key — the frontier models are just there, DeepSeek V4.1 Flash and Kimi K3 among them. Completely free, with no usage cap.
 LLM Models, Inference & Training JavaScript 37d 54363 - +1554 +13091
1020 YuJunZhiXue/dsh-purge ↗
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开，请查看署名要求部分
 LLM Models, Inference & Training JavaScript 73d 14581 - +708 +6460
887 Niko1221/Strata ↗
Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
 LLM Models, Inference & Training C++ 37d 37090 - +579 +5077
630 tashfeenahmed/freellmapi ↗
7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
 LLM Models, Inference & Training TypeScript 193d 45200 - +515 +4224
644 shiyu-coder/Kronos ↗
Kronos: A Foundation Model for the Language of Financial Markets
 LLM Models, Inference & Training Python 487d 44683 - +512 +1100
327 Alishahryar1/free-claude-code ↗
Use Claude Code, Codex, Antigravity, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
 LLM Models, Inference & Training Python 276d 65451 - +230 +2287
625 JustVugg/colibri ↗
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
 LLM Models, Inference & Training C 122d 45392 - +182 +1650
1004 deepseek-ai/DeepInfer ↗
DeepInfer: High-performance, large-scale inference framework for DeepSeek models
 LLM Models, Inference & Training Python 15d 24656 - +155 +2800
339 RVC-Boss/GPT-SoVITS ↗
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
 LLM Models, Inference & Training Python 1021d 64423 - +114 +841
981 google-research/timesfm ↗
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
 LLM Models, Inference & Training Python 915d 34944 - +100 +182
45 ollama/ollama ↗
Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
 LLM Models, Inference & Training Go 1223d 185479 - +97 +816
83 ggml-org/llama.cpp ↗
LLM inference in C/C++
 LLM Models, Inference & Training C++ 1331d 132992 - +94 +738
1229 SoMarkAI/OXR ↗
An open-source document parsing model with SOTA accuracy and speed, using an RSI loop during post-training.
 LLM Models, Inference & Training Python 26d 2761 - +91 -
712 lyogavin/airllm ↗
AirLLM 2.8T LLM inference with single 4GB GPU
 LLM Models, Inference & Training Jupyter Notebook 1237d 41434 - +86 +867
872 p-e-w/heretic ↗
Fully automatic censorship removal for language models
 LLM Models, Inference & Training Python 405d 37313 - +85 +899
1139 Folkbench/folkbench ↗
AI API 中转站评测排行与切换工具：实测可用率、延迟与价格 · Open-source AI API relay benchmark, rankings & switcher for Claude Code / Codex
 LLM Models, Inference & Training Rust 29d 4083 - +78 +504
1221 mmparko/dsh-purge ↗
DeepSeek Harness 破甲 (mirror of YuJunZhiXue/dsh-purge) — 上游即将删库，此为本仓库全量镜像：含全部分支/73 个 tag/完整提交历史。原作者署名与二开要求见 README 署名章节。
 LLM Models, Inference & Training JavaScript 18d 2828 - +77 +642
153 vllm-project/vllm ↗
A high-throughput and memory-efficient inference and serving engine for LLMs
 LLM Models, Inference & Training Python 1360d 96003 - +67 +659
632 chatanywhere/GPT_API_free ↗
免费大模型API，支持免费调用GPT、DeepSeek等主流模型，免费额度10000点，每日刷新！另付费价格最低官方1-2折！
 LLM Models, Inference & Training 1287d 45171 - +65 +376
1177 ranxianglei/billion-context ↗
A context-compression plugin for month-long single sessions (billions of tokens) ,small context windows (a 100K context is enough) and token savings (5x fewer tokens) .上下文压缩插件，月级长会话会话， 兼顾小窗口(100k上下文足矣)省token(省5倍token) . billion-context is all you need
 LLM Models, Inference & Training TypeScript 86d 3511 - +65 +879
1013 openai/math ↗
LLM Models, Inference & Training Lean 25d 17198 - +61 +1714
925 Tencent/WeKnora ↗
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
 LLM Models, Inference & Training Go 466d 36123 - +60 +813
733 OpenBMB/VoxCPM ↗
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
 LLM Models, Inference & Training Python 410d 40919 - +59 +399
988 NandhaKishorM/laya ↗
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
 LLM Models, Inference & Training Python 43d 34770 - +58 +565
227 unslothai/unsloth ↗
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
 LLM Models, Inference & Training Python 1067d 78878 - +51 +352
1276 OpenMOSS/MOSS-Image ↗
Official repository for MOSS-Image, a 5B-parameter base DiT model with high-quality text rendering, layout generation, and efficient fine-tuning.
 LLM Models, Inference & Training Python 15d 2357 - +50 -
401 karpathy/nanochat ↗
The best ChatGPT that $100 can buy.
 LLM Models, Inference & Training Python 383d 59179 - +49 +275
1281 mw00/project-maya ↗
Project Maya - run GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on your own GPU(s), from one card to 16, NVIDIA or AMD, Linux or Windows. Engine, server, dashboard, OpenAI- and Anthropic-compatible API, quant tools.
 LLM Models, Inference & Training Cuda 25d 2331 - +48 -
113 openai/whisper ↗
Robust Speech Recognition via Large-Scale Weak Supervision
 LLM Models, Inference & Training Python 1506d 111654 - +47 +451
364 BerriAI/litellm ↗
The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]
 LLM Models, Inference & Training Python 1193d 62455 - +47 +477
628 Wei-Shaw/sub2api ↗
Sub2API 一站式开源中转服务，让 Claude、Openai 、Gemini、Grok订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。
 LLM Models, Inference & Training Go 317d 45211 - +43 +437
503 QuantumNous/new-api ↗
Open-source AI gateway. Self-hosted LLM gateway with OpenAI-, Claude- and Gemini-compatible APIs, model routing, access control, usage tracking and cost accounting.
 LLM Models, Inference & Training Go 1086d 51102 - +42 +478
713 microsoft/BitNet ↗
Official inference framework for 1-bit LLMs
 LLM Models, Inference & Training C++ 817d 41419 - +41 +190
342 karpathy/nanoGPT ↗
The simplest, fastest repository for training/finetuning medium-sized GPTs.
 LLM Models, Inference & Training Python 1404d 64294 - +39 +237
337 jingyaogong/minimind ↗
🧠 Train a 64M-parameter LLM from scratch in just 2h!
 LLM Models, Inference & Training Python 826d 64692 - +27 +325
450 microsoft/VibeVoice ↗
Open-Source Frontier Voice AI
 LLM Models, Inference & Training Python 432d 55081 - +26 +127
52 huggingface/transformers ↗
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
LLM Models, Inference & Training Python 2924d 168399 - +24 +298
454 ggml-org/whisper.cpp ↗
Port of OpenAI's Whisper model in C/C++
 LLM Models, Inference & Training C++ 1497d 54830 - +24 +195
54 AUTOMATIC1111/stable-diffusion-webui ↗
Stable Diffusion web UI
 LLM Models, Inference & Training Python 1531d 165207 - +22 +80
527 mudler/LocalAI ↗
LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
 LLM Models, Inference & Training Go 1323d 50103 - +21 +168
629 janhq/jan ↗
Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.
 LLM Models, Inference & Training Rust 1171d 45204 - +21 +139
972 SillyTavern/SillyTavern ↗
LLM Frontend for Power Users.
 LLM Models, Inference & Training JavaScript 1360d 35159 - +20 +254
825 AlexsJones/llmfit ↗
Hundreds of models & providers. One command to find what runs on your hardware.
 LLM Models, Inference & Training Rust 258d 38524 - +19 +176
830 sgl-project/sglang ↗
SGLang is a high-performance serving framework for large language models and multimodal models.
 LLM Models, Inference & Training Python 1027d 38328 - +18 +525
128 pytorch/pytorch ↗
Tensors and Dynamic neural networks in Python with strong GPU acceleration
 LLM Models, Inference & Training Python 3731d 105047 - +17 0
559 exo-explore/exo ↗
Run frontier AI locally.
 LLM Models, Inference & Training Python 859d 48135 - +17 +103
31 tensorflow/tensorflow ↗
An Open Source Machine Learning Framework for Everyone
 LLM Models, Inference & Training C++ 4011d 201251 - +16 +21
1035 jaredpalmer/kev ↗
Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on your own
 LLM Models, Inference & Training Python 44d 9737 - +16 +228
1138 nykooi1/vibe-wise ↗
A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.
 LLM Models, Inference & Training Python 32d 4089 - +16 +171
772 VectifyAI/PageIndex ↗
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
 LLM Models, Inference & Training Python 578d 39931 - +15 +300

LLM Models, Inference & Training, movers in the last 7 days

#RepoCategory LanguageAge StarsNow24h7d
462 Ebony-Vinyl/dsh-our-free-model ↗
在 dsh 里装上这个插件即可，无需登录、注册或填 API Key，就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型——完全免费，不限量。 All you do is install this plugin in dsh: no login, no sign-up, no API key — the frontier models are just there, DeepSeek V4.1 Flash and Kimi K3 among them. Completely free, with no usage cap.
 LLM Models, Inference & Training JavaScript 37d 54363 - +1554 +13091
1020 YuJunZhiXue/dsh-purge ↗
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开，请查看署名要求部分
 LLM Models, Inference & Training JavaScript 73d 14581 - +708 +6460
887 Niko1221/Strata ↗
Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
 LLM Models, Inference & Training C++ 37d 37090 - +579 +5077
630 tashfeenahmed/freellmapi ↗
7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
 LLM Models, Inference & Training TypeScript 193d 45200 - +515 +4224
1004 deepseek-ai/DeepInfer ↗
DeepInfer: High-performance, large-scale inference framework for DeepSeek models
 LLM Models, Inference & Training Python 15d 24656 - +155 +2800
327 Alishahryar1/free-claude-code ↗
Use Claude Code, Codex, Antigravity, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
 LLM Models, Inference & Training Python 276d 65451 - +230 +2287
1013 openai/math ↗
LLM Models, Inference & Training Lean 25d 17198 - +61 +1714
625 JustVugg/colibri ↗
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
 LLM Models, Inference & Training C 122d 45392 - +182 +1650
644 shiyu-coder/Kronos ↗
Kronos: A Foundation Model for the Language of Financial Markets
 LLM Models, Inference & Training Python 487d 44683 - +512 +1100
872 p-e-w/heretic ↗
Fully automatic censorship removal for language models
 LLM Models, Inference & Training Python 405d 37313 - +85 +899
1177 ranxianglei/billion-context ↗
A context-compression plugin for month-long single sessions (billions of tokens) ,small context windows (a 100K context is enough) and token savings (5x fewer tokens) .上下文压缩插件，月级长会话会话， 兼顾小窗口(100k上下文足矣)省token(省5倍token) . billion-context is all you need
 LLM Models, Inference & Training TypeScript 86d 3511 - +65 +879
712 lyogavin/airllm ↗
AirLLM 2.8T LLM inference with single 4GB GPU
 LLM Models, Inference & Training Jupyter Notebook 1237d 41434 - +86 +867
339 RVC-Boss/GPT-SoVITS ↗
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
 LLM Models, Inference & Training Python 1021d 64423 - +114 +841
45 ollama/ollama ↗
Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
 LLM Models, Inference & Training Go 1223d 185479 - +97 +816
925 Tencent/WeKnora ↗
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
 LLM Models, Inference & Training Go 466d 36123 - +60 +813
83 ggml-org/llama.cpp ↗
LLM inference in C/C++
 LLM Models, Inference & Training C++ 1331d 132992 - +94 +738
153 vllm-project/vllm ↗
A high-throughput and memory-efficient inference and serving engine for LLMs
 LLM Models, Inference & Training Python 1360d 96003 - +67 +659
1221 mmparko/dsh-purge ↗
DeepSeek Harness 破甲 (mirror of YuJunZhiXue/dsh-purge) — 上游即将删库，此为本仓库全量镜像：含全部分支/73 个 tag/完整提交历史。原作者署名与二开要求见 README 署名章节。
 LLM Models, Inference & Training JavaScript 18d 2828 - +77 +642
988 NandhaKishorM/laya ↗
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
 LLM Models, Inference & Training Python 43d 34770 - +58 +565
830 sgl-project/sglang ↗
SGLang is a high-performance serving framework for large language models and multimodal models.
 LLM Models, Inference & Training Python 1027d 38328 - +18 +525
1139 Folkbench/folkbench ↗
AI API 中转站评测排行与切换工具：实测可用率、延迟与价格 · Open-source AI API relay benchmark, rankings & switcher for Claude Code / Codex
 LLM Models, Inference & Training Rust 29d 4083 - +78 +504
503 QuantumNous/new-api ↗
Open-source AI gateway. Self-hosted LLM gateway with OpenAI-, Claude- and Gemini-compatible APIs, model routing, access control, usage tracking and cost accounting.
 LLM Models, Inference & Training Go 1086d 51102 - +42 +478
364 BerriAI/litellm ↗
The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]
 LLM Models, Inference & Training Python 1193d 62455 - +47 +477
113 openai/whisper ↗
Robust Speech Recognition via Large-Scale Weak Supervision
 LLM Models, Inference & Training Python 1506d 111654 - +47 +451
628 Wei-Shaw/sub2api ↗
Sub2API 一站式开源中转服务，让 Claude、Openai 、Gemini、Grok订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。
 LLM Models, Inference & Training Go 317d 45211 - +43 +437
733 OpenBMB/VoxCPM ↗
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
 LLM Models, Inference & Training Python 410d 40919 - +59 +399
632 chatanywhere/GPT_API_free ↗
免费大模型API，支持免费调用GPT、DeepSeek等主流模型，免费额度10000点，每日刷新！另付费价格最低官方1-2折！
 LLM Models, Inference & Training 1287d 45171 - +65 +376
227 unslothai/unsloth ↗
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
 LLM Models, Inference & Training Python 1067d 78878 - +51 +352
337 jingyaogong/minimind ↗
🧠 Train a 64M-parameter LLM from scratch in just 2h!
 LLM Models, Inference & Training Python 826d 64692 - +27 +325
772 VectifyAI/PageIndex ↗
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
 LLM Models, Inference & Training Python 578d 39931 - +15 +300
52 huggingface/transformers ↗
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
LLM Models, Inference & Training Python 2924d 168399 - +24 +298
401 karpathy/nanochat ↗
The best ChatGPT that $100 can buy.
 LLM Models, Inference & Training Python 383d 59179 - +49 +275
972 SillyTavern/SillyTavern ↗
LLM Frontend for Power Users.
 LLM Models, Inference & Training JavaScript 1360d 35159 - +20 +254
1264 syv-ai/HyperQwen ↗
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.
 LLM Models, Inference & Training Python 77d 2463 - +14 +254
342 karpathy/nanoGPT ↗
The simplest, fastest repository for training/finetuning medium-sized GPTs.
 LLM Models, Inference & Training Python 1404d 64294 - +39 +237
1035 jaredpalmer/kev ↗
Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on your own
 LLM Models, Inference & Training Python 44d 9737 - +16 +228
454 ggml-org/whisper.cpp ↗
Port of OpenAI's Whisper model in C/C++
 LLM Models, Inference & Training C++ 1497d 54830 - +24 +195
793 google/langextract ↗
A Python library for extracting structured information from unstructured text using LLMs with precise source grounding and interactive visualization.
 LLM Models, Inference & Training Python 480d 39224 - +6 +191
713 microsoft/BitNet ↗
Official inference framework for 1-bit LLMs
 LLM Models, Inference & Training C++ 817d 41419 - +41 +190
981 google-research/timesfm ↗
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
 LLM Models, Inference & Training Python 915d 34944 - +100 +182
825 AlexsJones/llmfit ↗
Hundreds of models & providers. One command to find what runs on your hardware.
 LLM Models, Inference & Training Rust 258d 38524 - +19 +176
1138 nykooi1/vibe-wise ↗
A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.
 LLM Models, Inference & Training Python 32d 4089 - +16 +171
527 mudler/LocalAI ↗
LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
 LLM Models, Inference & Training Go 1323d 50103 - +21 +168
749 HKUDS/LightRAG ↗
[EMNLP2025] LightRAG: Simple and Fast Retrieval-Augmented Generation
 LLM Models, Inference & Training Python 759d 40454 - +10 +143
629 janhq/jan ↗
Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.
 LLM Models, Inference & Training Rust 1171d 45204 - +21 +139
801 RVC-Project/Retrieval-based-Voice-Conversion-WebUI ↗
Easily train a good VC model with voice data <= 10 mins!
 LLM Models, Inference & Training Python 1314d 38981 - +12 +138
450 microsoft/VibeVoice ↗
Open-Source Frontier Voice AI
 LLM Models, Inference & Training Python 432d 55081 - +26 +127
482 run-llama/llama_index ↗
LlamaIndex is the document processing platform for AI
 LLM Models, Inference & Training Python 1459d 52764 - +14 +120
559 exo-explore/exo ↗
Run frontier AI locally.
 LLM Models, Inference & Training Python 859d 48135 - +17 +103
471 lllyasviel/Fooocus ↗
Focus on prompting and generating
 LLM Models, Inference & Training Python 1179d 53777 - +8 +99

LLM Models, Inference & Training, breakouts over 24 hours

#RepoCategory LanguageAge StarsNow24h7d
1020 YuJunZhiXue/dsh-purge ↗
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开，请查看署名要求部分
 LLM Models, Inference & Training JavaScript 73d 14581 - +708 +6460
1229 SoMarkAI/OXR ↗
An open-source document parsing model with SOTA accuracy and speed, using an RSI loop during post-training.
 LLM Models, Inference & Training Python 26d 2761 - +91 -
462 Ebony-Vinyl/dsh-our-free-model ↗
在 dsh 里装上这个插件即可，无需登录、注册或填 API Key，就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型——完全免费，不限量。 All you do is install this plugin in dsh: no login, no sign-up, no API key — the frontier models are just there, DeepSeek V4.1 Flash and Kimi K3 among them. Completely free, with no usage cap.
 LLM Models, Inference & Training JavaScript 37d 54363 - +1554 +13091
1221 mmparko/dsh-purge ↗
DeepSeek Harness 破甲 (mirror of YuJunZhiXue/dsh-purge) — 上游即将删库，此为本仓库全量镜像：含全部分支/73 个 tag/完整提交历史。原作者署名与二开要求见 README 署名章节。
 LLM Models, Inference & Training JavaScript 18d 2828 - +77 +642
1276 OpenMOSS/MOSS-Image ↗
Official repository for MOSS-Image, a 5B-parameter base DiT model with high-quality text rendering, layout generation, and efficient fine-tuning.
 LLM Models, Inference & Training Python 15d 2357 - +50 -
1281 mw00/project-maya ↗
Project Maya - run GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on your own GPU(s), from one card to 16, NVIDIA or AMD, Linux or Windows. Engine, server, dashboard, OpenAI- and Anthropic-compatible API, quant tools.
 LLM Models, Inference & Training Cuda 25d 2331 - +48 -
1139 Folkbench/folkbench ↗
AI API 中转站评测排行与切换工具：实测可用率、延迟与价格 · Open-source AI API relay benchmark, rankings & switcher for Claude Code / Codex
 LLM Models, Inference & Training Rust 29d 4083 - +78 +504
1177 ranxianglei/billion-context ↗
A context-compression plugin for month-long single sessions (billions of tokens) ,small context windows (a 100K context is enough) and token savings (5x fewer tokens) .上下文压缩插件，月级长会话会话， 兼顾小窗口(100k上下文足矣)省token(省5倍token) . billion-context is all you need
 LLM Models, Inference & Training TypeScript 86d 3511 - +65 +879
887 Niko1221/Strata ↗
Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
 LLM Models, Inference & Training C++ 37d 37090 - +579 +5077
644 shiyu-coder/Kronos ↗
Kronos: A Foundation Model for the Language of Financial Markets
 LLM Models, Inference & Training Python 487d 44683 - +512 +1100
630 tashfeenahmed/freellmapi ↗
7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
 LLM Models, Inference & Training TypeScript 193d 45200 - +515 +4224
1004 deepseek-ai/DeepInfer ↗
DeepInfer: High-performance, large-scale inference framework for DeepSeek models
 LLM Models, Inference & Training Python 15d 24656 - +155 +2800
1264 syv-ai/HyperQwen ↗
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.
 LLM Models, Inference & Training Python 77d 2463 - +14 +254
625 JustVugg/colibri ↗
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
 LLM Models, Inference & Training C 122d 45392 - +182 +1650
1138 nykooi1/vibe-wise ↗
A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.
 LLM Models, Inference & Training Python 32d 4089 - +16 +171
1013 openai/math ↗
LLM Models, Inference & Training Lean 25d 17198 - +61 +1714
327 Alishahryar1/free-claude-code ↗
Use Claude Code, Codex, Antigravity, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
 LLM Models, Inference & Training Python 276d 65451 - +230 +2287
981 google-research/timesfm ↗
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
 LLM Models, Inference & Training Python 915d 34944 - +100 +182
872 p-e-w/heretic ↗
Fully automatic censorship removal for language models
 LLM Models, Inference & Training Python 405d 37313 - +85 +899
712 lyogavin/airllm ↗
AirLLM 2.8T LLM inference with single 4GB GPU
 LLM Models, Inference & Training Jupyter Notebook 1237d 41434 - +86 +867
1262 yibie/awesome-jev ↗
A curated list of public projects, integrations, and discussions built on Jev — TypeSafe AI's System One model for typed decisions.
 LLM Models, Inference & Training Python 44d 2518 - +5 +57
339 RVC-Boss/GPT-SoVITS ↗
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
 LLM Models, Inference & Training Python 1021d 64423 - +114 +841
988 NandhaKishorM/laya ↗
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
 LLM Models, Inference & Training Python 43d 34770 - +58 +565
925 Tencent/WeKnora ↗
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
 LLM Models, Inference & Training Go 466d 36123 - +60 +813
1035 jaredpalmer/kev ↗
Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on your own
 LLM Models, Inference & Training Python 44d 9737 - +16 +228
733 OpenBMB/VoxCPM ↗
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
 LLM Models, Inference & Training Python 410d 40919 - +59 +399
632 chatanywhere/GPT_API_free ↗
免费大模型API，支持免费调用GPT、DeepSeek等主流模型，免费额度10000点，每日刷新！另付费价格最低官方1-2折！
 LLM Models, Inference & Training 1287d 45171 - +65 +376
1196 Contrastive-LM/CLM ↗
Contrastive Language Models (CLMs) are a new class of System One model trained with a contrastive learning objective that connects states and actions.
 LLM Models, Inference & Training Python 38d 3140 - +4 +23
1059 mizorewww/laya-mlx ↗
Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API.
 LLM Models, Inference & Training Python 42d 7053 - +7 +34
713 microsoft/BitNet ↗
Official inference framework for 1-bit LLMs
 LLM Models, Inference & Training C++ 817d 41419 - +41 +190
628 Wei-Shaw/sub2api ↗
Sub2API 一站式开源中转服务，让 Claude、Openai 、Gemini、Grok订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。
 LLM Models, Inference & Training Go 317d 45211 - +43 +437
401 karpathy/nanochat ↗
The best ChatGPT that $100 can buy.
 LLM Models, Inference & Training Python 383d 59179 - +49 +275
503 QuantumNous/new-api ↗
Open-source AI gateway. Self-hosted LLM gateway with OpenAI-, Claude- and Gemini-compatible APIs, model routing, access control, usage tracking and cost accounting.
 LLM Models, Inference & Training Go 1086d 51102 - +42 +478
364 BerriAI/litellm ↗
The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]
 LLM Models, Inference & Training Python 1193d 62455 - +47 +477
83 ggml-org/llama.cpp ↗
LLM inference in C/C++
 LLM Models, Inference & Training C++ 1331d 132992 - +94 +738
153 vllm-project/vllm ↗
A high-throughput and memory-efficient inference and serving engine for LLMs
 LLM Models, Inference & Training Python 1360d 96003 - +67 +659
227 unslothai/unsloth ↗
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
 LLM Models, Inference & Training Python 1067d 78878 - +51 +352
342 karpathy/nanoGPT ↗
The simplest, fastest repository for training/finetuning medium-sized GPTs.
 LLM Models, Inference & Training Python 1404d 64294 - +39 +237
972 SillyTavern/SillyTavern ↗
LLM Frontend for Power Users.
 LLM Models, Inference & Training JavaScript 1360d 35159 - +20 +254
45 ollama/ollama ↗
Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
 LLM Models, Inference & Training Go 1223d 185479 - +97 +816
825 AlexsJones/llmfit ↗
Hundreds of models & providers. One command to find what runs on your hardware.
 LLM Models, Inference & Training Rust 258d 38524 - +19 +176
450 microsoft/VibeVoice ↗
Open-Source Frontier Voice AI
 LLM Models, Inference & Training Python 432d 55081 - +26 +127
830 sgl-project/sglang ↗
SGLang is a high-performance serving framework for large language models and multimodal models.
 LLM Models, Inference & Training Python 1027d 38328 - +18 +525
629 janhq/jan ↗
Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.
 LLM Models, Inference & Training Rust 1171d 45204 - +21 +139
454 ggml-org/whisper.cpp ↗
Port of OpenAI's Whisper model in C/C++
 LLM Models, Inference & Training C++ 1497d 54830 - +24 +195
113 openai/whisper ↗
Robust Speech Recognition via Large-Scale Weak Supervision
 LLM Models, Inference & Training Python 1506d 111654 - +47 +451
527 mudler/LocalAI ↗
LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
 LLM Models, Inference & Training Go 1323d 50103 - +21 +168
337 jingyaogong/minimind ↗
🧠 Train a 64M-parameter LLM from scratch in just 2h!
 LLM Models, Inference & Training Python 826d 64692 - +27 +325
863 google-ai-edge/mediapipe ↗
Cross-platform, customizable ML solutions for live and streaming media.
 LLM Models, Inference & Training C++ 2697d 37491 - +15 +96
907 jax-ml/jax ↗
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
 LLM Models, Inference & Training Python 2928d 36556 - +14 +64

LLM Models, Inference & Training, breakouts over 7 days

#RepoCategory LanguageAge StarsNow24h7d
1020 YuJunZhiXue/dsh-purge ↗
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开，请查看署名要求部分
 LLM Models, Inference & Training JavaScript 73d 14581 - +708 +6460
1177 ranxianglei/billion-context ↗
A context-compression plugin for month-long single sessions (billions of tokens) ,small context windows (a 100K context is enough) and token savings (5x fewer tokens) .上下文压缩插件，月级长会话会话， 兼顾小窗口(100k上下文足矣)省token(省5倍token) . billion-context is all you need
 LLM Models, Inference & Training TypeScript 86d 3511 - +65 +879
462 Ebony-Vinyl/dsh-our-free-model ↗
在 dsh 里装上这个插件即可，无需登录、注册或填 API Key，就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型——完全免费，不限量。 All you do is install this plugin in dsh: no login, no sign-up, no API key — the frontier models are just there, DeepSeek V4.1 Flash and Kimi K3 among them. Completely free, with no usage cap.
 LLM Models, Inference & Training JavaScript 37d 54363 - +1554 +13091
1221 mmparko/dsh-purge ↗
DeepSeek Harness 破甲 (mirror of YuJunZhiXue/dsh-purge) — 上游即将删库，此为本仓库全量镜像：含全部分支/73 个 tag/完整提交历史。原作者署名与二开要求见 README 署名章节。
 LLM Models, Inference & Training JavaScript 18d 2828 - +77 +642
887 Niko1221/Strata ↗
Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
 LLM Models, Inference & Training C++ 37d 37090 - +579 +5077
1139 Folkbench/folkbench ↗
AI API 中转站评测排行与切换工具：实测可用率、延迟与价格 · Open-source AI API relay benchmark, rankings & switcher for Claude Code / Codex
 LLM Models, Inference & Training Rust 29d 4083 - +78 +504
1004 deepseek-ai/DeepInfer ↗
DeepInfer: High-performance, large-scale inference framework for DeepSeek models
 LLM Models, Inference & Training Python 15d 24656 - +155 +2800
1264 syv-ai/HyperQwen ↗
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.
 LLM Models, Inference & Training Python 77d 2463 - +14 +254
1013 openai/math ↗
LLM Models, Inference & Training Lean 25d 17198 - +61 +1714
630 tashfeenahmed/freellmapi ↗
7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
 LLM Models, Inference & Training TypeScript 193d 45200 - +515 +4224
1138 nykooi1/vibe-wise ↗
A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.
 LLM Models, Inference & Training Python 32d 4089 - +16 +171
625 JustVugg/colibri ↗
Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
 LLM Models, Inference & Training C 122d 45392 - +182 +1650
327 Alishahryar1/free-claude-code ↗
Use Claude Code, Codex, Antigravity, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
 LLM Models, Inference & Training Python 276d 65451 - +230 +2287
644 shiyu-coder/Kronos ↗
Kronos: A Foundation Model for the Language of Financial Markets
 LLM Models, Inference & Training Python 487d 44683 - +512 +1100
872 p-e-w/heretic ↗
Fully automatic censorship removal for language models
 LLM Models, Inference & Training Python 405d 37313 - +85 +899
1035 jaredpalmer/kev ↗
Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on your own
 LLM Models, Inference & Training Python 44d 9737 - +16 +228
1262 yibie/awesome-jev ↗
A curated list of public projects, integrations, and discussions built on Jev — TypeSafe AI's System One model for typed decisions.
 LLM Models, Inference & Training Python 44d 2518 - +5 +57
925 Tencent/WeKnora ↗
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
 LLM Models, Inference & Training Go 466d 36123 - +60 +813
712 lyogavin/airllm ↗
AirLLM 2.8T LLM inference with single 4GB GPU
 LLM Models, Inference & Training Jupyter Notebook 1237d 41434 - +86 +867
988 NandhaKishorM/laya ↗
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
 LLM Models, Inference & Training Python 43d 34770 - +58 +565
830 sgl-project/sglang ↗
SGLang is a high-performance serving framework for large language models and multimodal models.
 LLM Models, Inference & Training Python 1027d 38328 - +18 +525
339 RVC-Boss/GPT-SoVITS ↗
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
 LLM Models, Inference & Training Python 1021d 64423 - +114 +841
733 OpenBMB/VoxCPM ↗
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
 LLM Models, Inference & Training Python 410d 40919 - +59 +399
628 Wei-Shaw/sub2api ↗
Sub2API 一站式开源中转服务，让 Claude、Openai 、Gemini、Grok订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。
 LLM Models, Inference & Training Go 317d 45211 - +43 +437
503 QuantumNous/new-api ↗
Open-source AI gateway. Self-hosted LLM gateway with OpenAI-, Claude- and Gemini-compatible APIs, model routing, access control, usage tracking and cost accounting.
 LLM Models, Inference & Training Go 1086d 51102 - +42 +478
632 chatanywhere/GPT_API_free ↗
免费大模型API，支持免费调用GPT、DeepSeek等主流模型，免费额度10000点，每日刷新！另付费价格最低官方1-2折！
 LLM Models, Inference & Training 1287d 45171 - +65 +376
364 BerriAI/litellm ↗
The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]
 LLM Models, Inference & Training Python 1193d 62455 - +47 +477
772 VectifyAI/PageIndex ↗
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
 LLM Models, Inference & Training Python 578d 39931 - +15 +300
1196 Contrastive-LM/CLM ↗
Contrastive Language Models (CLMs) are a new class of System One model trained with a contrastive learning objective that connects states and actions.
 LLM Models, Inference & Training Python 38d 3140 - +4 +23
972 SillyTavern/SillyTavern ↗
LLM Frontend for Power Users.
 LLM Models, Inference & Training JavaScript 1360d 35159 - +20 +254
153 vllm-project/vllm ↗
A high-throughput and memory-efficient inference and serving engine for LLMs
 LLM Models, Inference & Training Python 1360d 96003 - +67 +659
83 ggml-org/llama.cpp ↗
LLM inference in C/C++
 LLM Models, Inference & Training C++ 1331d 132992 - +94 +738
981 google-research/timesfm ↗
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
 LLM Models, Inference & Training Python 915d 34944 - +100 +182
337 jingyaogong/minimind ↗
🧠 Train a 64M-parameter LLM from scratch in just 2h!
 LLM Models, Inference & Training Python 826d 64692 - +27 +325
793 google/langextract ↗
A Python library for extracting structured information from unstructured text using LLMs with precise source grounding and interactive visualization.
 LLM Models, Inference & Training Python 480d 39224 - +6 +191
1059 mizorewww/laya-mlx ↗
Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API.
 LLM Models, Inference & Training Python 42d 7053 - +7 +34
401 karpathy/nanochat ↗
The best ChatGPT that $100 can buy.
 LLM Models, Inference & Training Python 383d 59179 - +49 +275
713 microsoft/BitNet ↗
Official inference framework for 1-bit LLMs
 LLM Models, Inference & Training C++ 817d 41419 - +41 +190
825 AlexsJones/llmfit ↗
Hundreds of models & providers. One command to find what runs on your hardware.
 LLM Models, Inference & Training Rust 258d 38524 - +19 +176
227 unslothai/unsloth ↗
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
 LLM Models, Inference & Training Python 1067d 78878 - +51 +352
45 ollama/ollama ↗
Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
 LLM Models, Inference & Training Go 1223d 185479 - +97 +816
113 openai/whisper ↗
Robust Speech Recognition via Large-Scale Weak Supervision
 LLM Models, Inference & Training Python 1506d 111654 - +47 +451
342 karpathy/nanoGPT ↗
The simplest, fastest repository for training/finetuning medium-sized GPTs.
 LLM Models, Inference & Training Python 1404d 64294 - +39 +237
454 ggml-org/whisper.cpp ↗
Port of OpenAI's Whisper model in C/C++
 LLM Models, Inference & Training C++ 1497d 54830 - +24 +195
801 RVC-Project/Retrieval-based-Voice-Conversion-WebUI ↗
Easily train a good VC model with voice data <= 10 mins!
 LLM Models, Inference & Training Python 1314d 38981 - +12 +138
749 HKUDS/LightRAG ↗
[EMNLP2025] LightRAG: Simple and Fast Retrieval-Augmented Generation
 LLM Models, Inference & Training Python 759d 40454 - +10 +143
527 mudler/LocalAI ↗
LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
 LLM Models, Inference & Training Go 1323d 50103 - +21 +168
629 janhq/jan ↗
Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.
 LLM Models, Inference & Training Rust 1171d 45204 - +21 +139
863 google-ai-edge/mediapipe ↗
Cross-platform, customizable ML solutions for live and streaming media.
 LLM Models, Inference & Training C++ 2697d 37491 - +15 +96
810 stanfordnlp/dspy ↗
DSPy: The framework for programming—not prompting—language models
 LLM Models, Inference & Training Python 1391d 38868 - +8 +95

LLM Models, Inference & Training, new and rising

#RepoCategory LanguageAge StarsNow24h7d
462 Ebony-Vinyl/dsh-our-free-model ↗
在 dsh 里装上这个插件即可，无需登录、注册或填 API Key，就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型——完全免费，不限量。 All you do is install this plugin in dsh: no login, no sign-up, no API key — the frontier models are just there, DeepSeek V4.1 Flash and Kimi K3 among them. Completely free, with no usage cap.
 LLM Models, Inference & Training JavaScript 37d 54363 - +1554 +13091
887 Niko1221/Strata ↗
Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
 LLM Models, Inference & Training C++ 37d 37090 - +579 +5077
988 NandhaKishorM/laya ↗
Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
 LLM Models, Inference & Training Python 43d 34770 - +58 +565
1004 deepseek-ai/DeepInfer ↗
DeepInfer: High-performance, large-scale inference framework for DeepSeek models
 LLM Models, Inference & Training Python 15d 24656 - +155 +2800
1013 openai/math ↗
LLM Models, Inference & Training Lean 25d 17198 - +61 +1714
1020 YuJunZhiXue/dsh-purge ↗
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开，请查看署名要求部分
 LLM Models, Inference & Training JavaScript 73d 14581 - +708 +6460
1029 pathwaycom/arc-task-gen ↗
Generates original ARC-AGI-1-style tasks distribution-matched to the public eval set.
 LLM Models, Inference & Training Python 88d 10934 - -7 -36
1035 jaredpalmer/kev ↗
Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on your own
 LLM Models, Inference & Training Python 44d 9737 - +16 +228
1059 mizorewww/laya-mlx ↗
Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API.
 LLM Models, Inference & Training Python 42d 7053 - +7 +34
1138 nykooi1/vibe-wise ↗
A Claude Code / Codex plugin that helps you learn how to build while AI writes the code.
 LLM Models, Inference & Training Python 32d 4089 - +16 +171
1139 Folkbench/folkbench ↗
AI API 中转站评测排行与切换工具：实测可用率、延迟与价格 · Open-source AI API relay benchmark, rankings & switcher for Claude Code / Codex
 LLM Models, Inference & Training Rust 29d 4083 - +78 +504
1177 ranxianglei/billion-context ↗
A context-compression plugin for month-long single sessions (billions of tokens) ,small context windows (a 100K context is enough) and token savings (5x fewer tokens) .上下文压缩插件，月级长会话会话， 兼顾小窗口(100k上下文足矣)省token(省5倍token) . billion-context is all you need
 LLM Models, Inference & Training TypeScript 86d 3511 - +65 +879
1196 Contrastive-LM/CLM ↗
Contrastive Language Models (CLMs) are a new class of System One model trained with a contrastive learning objective that connects states and actions.
 LLM Models, Inference & Training Python 38d 3140 - +4 +23
1202 Taichu-AI/ZDTaichu5.0-9B ↗
LLM Models, Inference & Training Python 57d 3063 - -22 -62
1221 mmparko/dsh-purge ↗
DeepSeek Harness 破甲 (mirror of YuJunZhiXue/dsh-purge) — 上游即将删库，此为本仓库全量镜像：含全部分支/73 个 tag/完整提交历史。原作者署名与二开要求见 README 署名章节。
 LLM Models, Inference & Training JavaScript 18d 2828 - +77 +642
1229 SoMarkAI/OXR ↗
An open-source document parsing model with SOTA accuracy and speed, using an RSI loop during post-training.
 LLM Models, Inference & Training Python 26d 2761 - +91 -
1262 yibie/awesome-jev ↗
A curated list of public projects, integrations, and discussions built on Jev — TypeSafe AI's System One model for typed decisions.
 LLM Models, Inference & Training Python 44d 2518 - +5 +57
1264 syv-ai/HyperQwen ↗
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.
 LLM Models, Inference & Training Python 77d 2463 - +14 +254
1276 OpenMOSS/MOSS-Image ↗
Official repository for MOSS-Image, a 5B-parameter base DiT model with high-quality text rendering, layout generation, and efficient fine-tuning.
 LLM Models, Inference & Training Python 15d 2357 - +50 -
1281 mw00/project-maya ↗
Project Maya - run GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on your own GPU(s), from one card to 16, NVIDIA or AMD, Linux or Windows. Engine, server, dashboard, OpenAI- and Anthropic-compatible API, quant tools.
 LLM Models, Inference & Training Cuda 25d 2331 - +48 -