OpenRouter Update
Sakana Fugu Ultra ($5/M tokens prompt) has arrived on OpenRouter with a massive 1M context window. Fugu Ultra is Sakana’s “one model to command them all” orchestrator, designed to coordinate multiple specialized models across complex multi-step tasks.
Limited free models expiring June 29 (3 days left):
- Anthropic: Claude Opus 4.6 (Fast) — $30/M tokens prompt, $150/M tokens completion, 1M context
- StepFun: Step 3.5 Flash — $0.09/M tokens prompt, $0.30/M tokens completion, 262K context
Top GitHub Repos
yo-WASSUP/Good-Badminton ⭐523 — An AI-powered badminton Hawk-Eye system using computer vision for line-calling and shot tracking. Python.
benchflow-ai/awesome-evals ⭐325 — A curated library of the best resources for building and evaluating AI agents — papers, blog posts, and tools.
trotsky1997/OpenFugu ⭐234 — Open-source reimplementation of Sakana Fugu, the “one model to command them all” LLM orchestrator. Python.
iart-ai/motion-skills ⭐209 — 50 open-source skills that teach AI coding agents to make motion graphics, animation, and video. HTML.
i5ting/learn-ai-practice ⭐182 — A learning resource for AI practice.
New Hugging Face Models
| Model | Type | Downloads |
|---|---|---|
| woh7539/omx_act_policy | Robotics ACT policy (LeRobot) | 85 |
| RikZD/ZNX-VL-Coder-7B | Vision-language coding model | 0 |
| Ritesh-hf/Ask-Solve-Generate-VARGPT-v1.1 | Multimodal VQA + image generation | 0 |
| csukuangfj/kaldifeat | Audio feature extraction | 0 |
| gdiamos/relm-3-e2b-it | Gemma 4 based RL fine-tune | 1 |
| aimalias/k33l3y | Text-to-image LoRA (Diffusers) | 8 |
| gguf-org/krea-2-gguf | Krea 2 GGUF quantized | 0 |
| KissTheHabit/IDA_AI | IDA lattice transformers | 28 |
| mulemp/EmilyMars | — | 0 |
| zoarag/SynthiaAI | — | 0 |
Standouts: The woh7539/omx_act_policy model brings a robotics ACT policy to Hugging Face via LeRobot, suggesting growing interest in open-source robotics. gguf-org/krea-2-gguf provides a GGUF-quantized version of Krea’s image model, making it accessible for local inference via llama.cpp. gdiamos/relm-3-e2b-it continues the Gemma 4 fine-tuning trend with RL-based instruction tuning.