AI Tools Radar › Categories › Models & research
AI models & research code
Model releases, open weights, versioned model families (Qwen, DeepSeek, Llama, Gemma, Mistral) and paper or official research code. 36 items in today's radar (data fetched 21 Oct 2026, 14:13 (Berlin time)). Categories are assigned by fixed keyword rules on GitHub topics, repo names, descriptions and Show HN titles; an item can be in several categories. How categories work.
All 393Agents & automation 196Coding assistants & dev tools 168Chat apps & assistants 28LLM inference & serving 36RAG & search 28Training & fine-tuning 11Models & research 36Image, video & audio generation 52Voice & speech 21Computer vision 8MCP, skills & plugins 146Evaluation & observability 22Data & datasets 30Local / on-device AI 45Security & sandboxing 33Learning & resources 48Other 16
Trending on GitHub today (8)
| Repository | Stars today | Total stars | Language | License | Created |
|---|---|---|---|---|---|
| 1. Comfy-Org/ComfyUI The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world. | 1,206 | 141,213 | Python | GPL-3.0 | 2023-01-17 |
| 2. ollama/ollama Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models. | 437 | 184,129 | Go | MIT | 2023-06-26 |
| 3. lyogavin/airllm AirLLM 2.8T LLM inference with single 4GB GPU | 386 | 40,040 | Jupyter Notebook | Apache-2.0 | 2023-06-12 |
| 4. vllm-project/vllm A high-throughput and memory-efficient inference and serving engine for LLMs | 128 | 94,666 | Python | Apache-2.0 | 2023-02-09 |
| 5. mudler/LocalAI LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required. | 126 | 49,823 | Go | MIT | 2023-03-18 |
| 6. google-research/google-research Google Research | 27 | 38,995 | Jupyter Notebook | Apache-2.0 | 2018-10-04 |
| 7. HandsOnLLM/Hands-On-Large-Language-Models Official code repo for the O'Reilly Book - "Hands-On Large Language Models" | 11 | 29,801 | Jupyter Notebook | Apache-2.0 | 2024-06-28 |
| 8. facebookresearch/sam2 The repository provides code for running inference with the Meta Segment Anything Model 2 (SAM 2), links for downloading the trained model checkpoints, and example notebooks that show how to use the model. | 10 | 20,032 | Jupyter Notebook | Apache-2.0 | 2024-07-29 |
Trending on GitHub this week (10)
| Repository | Stars this week | Total stars | Language | License | Created |
|---|---|---|---|---|---|
| 1. Comfy-Org/ComfyUI The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world. | 3,364 | 141,213 | Python | GPL-3.0 | 2023-01-17 |
| 2. lyogavin/airllm AirLLM 2.8T LLM inference with single 4GB GPU | 1,694 | 40,040 | Jupyter Notebook | Apache-2.0 | 2023-06-12 |
| 3. ollama/ollama Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models. | 1,391 | 184,129 | Go | MIT | 2023-06-26 |
| 4. lightseekorg/tokenspeed TokenSpeed is a speed-of-light LLM inference engine. | 503 | 2,737 | Python | MIT | 2026-05-06 |
| 5. mudler/LocalAI LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required. | 300 | 49,823 | Go | MIT | 2023-03-18 |
| 6. HandsOnLLM/Hands-On-Large-Language-Models Official code repo for the O'Reilly Book - "Hands-On Large Language Models" | 229 | 29,801 | Jupyter Notebook | Apache-2.0 | 2024-06-28 |
| 7. google-research/google-research Google Research | 86 | 38,995 | Jupyter Notebook | Apache-2.0 | 2018-10-04 |
| 8. facebookresearch/segment-anything The repository provides code for running inference with the SegmentAnything Model (SAM), links for downloading the trained model checkpoints, and example notebooks that show how to use the model. | 68 | 55,039 | Jupyter Notebook | Apache-2.0 | 2023-03-23 |
| 9. CompVis/stable-diffusion A latent text-to-image diffusion model | 54 | 73,530 | Jupyter Notebook | Other | 2022-08-10 |
| 10. facebookresearch/sam2 The repository provides code for running inference with the Meta Segment Anything Model 2 (SAM 2), links for downloading the trained model checkpoints, and example notebooks that show how to use the model. | 39 | 20,032 | Jupyter Notebook | Apache-2.0 | 2024-07-29 |
New repos of the last 30 days (23)
| Repository | Age | Total stars | Language | License | Created |
|---|---|---|---|---|---|
| 1. 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. | 26 d | 32,653 | JavaScript | MIT | 2026-09-24 |
| 2. 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. | 26 d | 29,588 | C++ | MIT | 2026-09-24 |
| 3. yetone/magpie Every agent's model. One place. Codex on DeepSeek, Claude Code on Kimi, from the menu bar. | 27 d | 18,590 | Go | MIT | 2026-09-23 |
| 4. Arekejoker/qwen21-nsfw-pose-skill Qwen-Image 2.1 NSFW/姿势出图 skill：作者级提示词配方 + NSFW LoRA + 136 姿势 VNCCS 人偶控制；适用 Claude Code / Codex / Cursor / OpenClaw | 8 d | 3,556 | Python | Apache-2.0 | 2026-10-13 |
| 5. mw00/project-maya Project Maya - run GLM-5.3-Flash (321B MoE) and DeepSeek-V4.1-Flash-1017 (748B 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. | 14 d | 1,350 | C++ | MIT | 2026-10-06 |
| 6. StayLameBro/backburner Your iPhone helps your Mac run a 27B model: faster prompt reading and more context over a USB-C cable | 19 d | 1,092 | Python | MIT | 2026-10-01 |
| 7. StartLuxLabs/StartLux-Decision Open decision models from 0.8B to 27B: a calibrated probability for every option in one pass, and a short thought only when unsure. Engine for vLLM, llama.cpp and MLX. | 22 d | 1,073 | Python | Apache-2.0 | 2026-09-29 |
| 8. Harvil1/codeAgent CodeAgent is an open-source coding agent that lives in your terminal. It understands your repository, plans multi-step changes, writes and edits files, runs commands and tests, and iterates until the task is done — all with a transparent, extensible tool loop. | 19 d | 1,026 | Python | MIT | 2026-10-02 |
| 9. angel291592/Intent-Router Intent compiler for AI agents — converges vague requests into typed IntentSpec contracts (probe, ask, or halt before routing), the input layer for routers and typed-decision models like Jev & Laya | 29 d | 718 | Python | MIT | 2026-09-22 |
| 10. strands-labs/strands-decider A small, fast decision model, or system one model, for agentic workflows. Pick between options or rate on a scale faster than an LLM, with a calibrated confidence on every decision. | 21 d | 684 | Python | Apache-2.0 | 2026-09-29 |
| 11. cobanov/PS5LM llama.cpp on a jailbroken PS5: run any GGUF model natively on the console, starting with Qwen 3.8 | 19 d | 621 | C++ | GPL-3.0 | 2026-10-01 |
| 12. kydlikebtc/awesome-jev 1207 public resources for Jev, TypeSafe AI's System One decision model, indexed by decision pattern. Source citations, dated link checks and scheduled call-site text checks; runtime and performance are not independently tested here. EN/中文, JSON schema and platform compatibility. | 29 d | 593 | Python | Other | 2026-09-22 |
| 13. UnderleveledBuilder/awesome-short-drama-skills AI 短剧 · 漫剧 Skill（技能包）精选：剧本、改编、分镜、Seedance / 可灵 / MiniMax 视频提示词，千问、豆包、龙虾 OpenClaw、WorkBuddy、Claude Code、Codex 等智能体可用 | Curated agent skills for AI short drama, micro drama / microdrama, vertical drama and manju (AI comic drama): screenwriting, storyboards, Seedance / Kling / MiniMax video prompts | 7 d | 576 | HTML | CC0-1.0 | 2026-10-13 |
| 14. malevrigns/agent-jev AgentJev-0.6B - a fast 'System One' decision model for AI Agents: feed it any unstructured state (diffs, traces, logs) and structured questions, get calibrated probability distributions back in one ~50ms forward pass. Zero output-token decoding. | 29 d | 394 | Python | Apache-2.0 | 2026-09-21 |
| 15. VecSzn/PolyStrata Run large AI models on your own PC: Qwen3.8-Flash-Next on Strata's engine, and GLM 5.3 Flash, DeepSeek V4 Flash, LFM2.5, Qwen3.6, Ornith 1.5, Gemma 4, Tiel Coder and MiniCPM-V on a second engine written for PolyStrata. One-click install for Windows and Linux, OpenAI/Anthropic API on localhost. | 14 d | 324 | C++ | MIT | 2026-10-06 |
| 16. mikamikasuki/forest FOREST is a research agent for long-horizon science, where evidence gets lost, errors compound, and goals drift. It explores ideas, reads literature, forms hypotheses, runs real experiments, analyzes results, revises paths, and builds a full paper—all in an editable graph you can inspect, branch, redirect, with claims and figures linked to evidence | 22 d | 308 | Python | Apache-2.0 | 2026-09-29 |
| 17. Atoll92/agent-standup Your Claude Code, Codex and Mistral Vibe sessions as a live, animated team call: who is working, who waits on whom, who needs you. VS Code extension, terminal command, desktop app. Runs locally, zero tokens. | 11 d | 300 | HTML | MIT | 2026-10-10 |
| 18. pinganbbOvO/ai-companion-kit 给你的 AI 一副身体，和一点人味 —— 会记事、会看场合的 QQ 陪聊 + 能控制安卓手机的 MCP | 13 d | 281 | JavaScript | MIT | 2026-10-08 |
| 19. lokicorvus/crosspet 跨 AI 桌宠：跟着你用的 AI 换形象、跟着 AI 的工作状态做动作 | 17 d | 268 | C# | Other | 2026-10-03 |
| 20. ANDRETRIPOL/OpenGhost Open desktop AI agent for Windows, macOS and Linux, written from scratch: its own agent engine, its own browser tooling and its own drawing engine that shows what it explains. | 25 d | 266 | JavaScript | MIT | 2026-09-25 |
| 21. Indras-Mirror/Xolotl Xolotl (formerly Strata-DSV4.1): DeepSeek-V4.1-Flash, all 552B, on one gaming PC (RTX 4090 + 90 GB RAM) at ~27 tok/s, plus GLM-5.3-Flash, Qwen3.8-Flash-Next and Qwen3.8-27B. A heavily extended fork of Strata: MoE experts tiered across VRAM, pinned RAM and NVMe; OpenAI + Anthropic API. | 14 d | 262 | C++ | MIT | 2026-10-06 |
| 22. MemTensor/MemT-1.0 Official repository for MemT-1.0 | 2 d | 251 | Python | Other | 2026-10-19 |
| 23. raullenchai/esp32-thinks A multi-model Muse: personal AI voice assistant on ESP32 - wake word or knob, any OpenAI-compatible brain (STT/chat/TTS mix-and-match), streaming replies, out of the box. Built on muse-gadget-sdk. | 8 d | 236 | Shell | MIT | 2026-10-12 |
AI launches on Show HN (2)
| Show HN post | Points | Comments | Posted |
|---|---|---|---|
| 1. Show HN: Qwen3-4B with the AI writing tics trained out HN discussion · by erikkannike | 155 | 111 | 2026-10-19 |
| 2. Show HN: Jevman – AI decision models play Pac-Man HN discussion · by felix089 | 82 | 23 | 2026-10-08 |