An open-source proxy that stacks 20+ inference-time reasoning techniques onto any OpenAI-compatible endpoint, trading extra API calls for large accuracy gains.
A unified framework wraps nine robot simulators behind one API, pools 30+ datasets, and standardizes benchmarks for both imitation and reinforcement learning.
Pi is a minimal, aggressively extensible coding agent harness with unified access to 15+ LLM providers, tree-structured sessions, and self-modifying customization.
A lightweight Python library pulls YouTube captions directly without Selenium, API keys, or headless browsers, and ships with proxy and translation support.
A community fine-tune of Alibaba's Qwen3.6 MoE hits 161.9 tokens per second on a single RTX 5090, with a 262k context window and one-shot frontend generation.
RAG-Anything is an open-source, graph-powered RAG framework from HKUDS that processes text, images, tables, and equations inside a single unified pipeline — no more stitching together a dozen specialized tools.