OpenAI published an open-source plugin that lets Claude Code users invoke Codex for reviews, background tasks, and session handoffs with slash commands.
An open-weights Mixture-of-Experts model gets fine-tuned on 8,000 Claude Opus 4.7 reasoning traces, porting frontier chain-of-thought style into Apache 2.0 weights.
A new optimization framework shows convergence speed depends on the local geometry an algorithm actually traverses, not worst-case bounds over the entire space.
A community-built terminal coding agent wired to xAI's Grok API, with sub-agents, sandboxed verification, Telegram remote control, and X search built in.
A new framework treats 3D scene layout as an autoregressive geometry problem, hitting 19% better physical plausibility and 65% faster inference than the prior state of the art.
Hugging Face's Rust-based embeddings server delivers up to 3x throughput over ONNX with dynamic batching, Flash Attention, and support for Qwen3, ModernBERT, and Gemma3.
A new framework trains a 11.5-billion parameter atomistic foundation model on two exascale supercomputers, hitting 1.2 EFLOPS and cutting weeks of training down to hours.