AI Briefing#llm
The latest AI and developer news about #llm, with the original source and a short summary.
The latest AI and developer news about #llm, with the original source and a short summary.
Firecrawl released anydoc, a Rust library that converts 14 document formats into Markdown.
Meta has released a terminal coding agent with session-based sub-agents and a dedicated model.
Oracle has prohibited contributions of AI-generated code and content to OpenJDK.
Hermes Agent now supports Vercel AI Gateway and Vercel Sandbox.
Cloudflare has launched the beta version of Radar Researcher, which analyzes internet data through natural language questions.
Moonshot AI joins the competition by releasing Kimi K3 as an open-weight model.
Liquid AI released a 2.6B agent model capable of running on mobile phones and CPUs.
The release date for the 2.4T model known as Qwen3.8-Max has been set for next Wednesday.
Ant Group has released Ling-3.0-flash, a 124B model, under an unrestricted MIT license.
DeepSeek has released the official version of V4-Flash, which supports a 1 million token context window.
The Rust project has formalized rules for LLM usage in the rust-lang/rust contribution process.
In AI agent development, RL environments serve as the core foundation for leveraging data and compute.
DSPy's Flex leverages GEPA to automatically optimize not just prompts but also program code.
DeepAmbigQA combines ambiguity and multi-hop reasoning to evaluate the answer completeness of LLMs.
FuriosaAI presented four papers at ICML 2026 addressing efficiency in LLM inference, training, and serving.
DLR-Lock makes unauthorized modification of pre-trained weights difficult while maintaining model performance.
Hugging Face has added Baseten as an Inference Provider to support serverless model inference.
The Qwen development team teased the launch of a 27B model and key specifications for upcoming models.
AI Gateway can now be purchased via AWS Marketplace and integrated into AWS billing.
InternLM has released Intern S2 Mobius, a derivative model of Qwen3.5-35B.
Codex sent a request of approximately 9,435 tokens to the model for an input of just 16 characters.
LLM 0.32 supports reasoning traces, server-side tools, and OpenAI-compatible endpoints.
AMD has released a fully open 16B MoE model trained on ROCm.
AFM3 20B is a MoE model that selects only about 20% of MLP layers to activate per prompt.