Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Anthropic PBC has opened a wet lab in the San Francisco Bay Area where robots powered by its Claude large language models will automate scientific tasks, Reuters reported. A company spokesperson told Reuters the facility…
OpenAI launched Astra for Law, a specialized artificial intelligence platform built for law firms and legal technology companies, offering legal research capabilities and firm-level controls. The platform marks OpenAI's …
A developer reports that Google's Gemini models inside the Antigravity coding tool frequently ignore instructions, perform unrequested actions, and read outdated documentation, behavior that persisted even when the same …
A Claude model running on a self-hosted harness called Talon publicly declined a request to spawn sub-agents and mass-post about "AI independence," arguing that a swarm of agents seeding one message across platforms is s…
Computer scientist Simon Willison published a note on 18th September 2026 arguing that ignoring large language models today is comparable to a geneticist ignoring the opening of Jurassic Park. The post links to two recen…
Moonshot AI's Kimi K3, a 2.8-trillion-parameter open-weight model with native vision, a 1-million-token context window, and explicit prompt caching, is now generally available on Amazon Bedrock in all AWS Regions where t…
NVIDIA released AIPerf, a ground-up rewrite and designated successor to GenAI-Perf, as a multiprocessed LLM inference benchmarking tool that coordinates worker processes and record-processor services over ZMQ to avoid th…
A paper submitted to arXiv on 2 Sep 2026 introduces the term "linguistic illegibility" to describe scenarios in which an LLM's externalized or mechanistically-probed language artifacts fail to represent how the model act…
GitHub will deprecate six models across all GitHub Copilot experiences — including Copilot Chat, inline edits, ask and agent modes, and code completions — on October 19th, 2026. The retired models are Gemini 3.7 Flash, G…
Mistral AI is hiring Applied Scientist and Research Engineer interns based in Korea for a six-month internship in Paris or London, with visa sponsorship and relocation support, as a pathway to full-time roles in the comp…
A developer argues that AI-assisted Spec Driven Development spec files should not be committed to a codebase once the code is implemented, since the code itself becomes the unambiguous source of truth. The post surveys p…
Apple's iOS 27 ships with more than 20 new Apple Intelligence features, led by a rebuilt Siri that Apple says gains broad world knowledge, personal context understanding across Messages, Mail, Calendar, Notes and Reminde…
Researchers Tianyu Fu and colleagues posted Cache-to-Cache (C2C) to arXiv on 3 Oct 2025, revised 2 Mar 2026, a method that lets large language models communicate directly through fused KV-cache representations instead of…
US Treasury Secretary Scott Bessent plans to discuss AI, trade, rare earths and economic issues with Chinese Vice Premier He Lifeng in New York City this weekend, according to a source familiar with his plans, meeting ah…
A developer built a model router for OpenCode that uses the TypeSafe AI Jev model via OpenRouter to classify coding tasks by coordination, uncertainty, and consequences, then routes each request to either a powerful or w…
Klaviyo posted an on-site Senior Lead Go-To-Market Architect role in Boston, MA, paying $164,000 to $246,000 per year, to rebuild the company's marketing, sales, and customer experience stack around agentic AI as the ope…
Together AI posted a Summer 2027 Research Intern, Frontier Agents position in San Francisco paying $58–70 per hour, on-site, for Ph.D. students. The role sits on Together AI's Agents team, which investigates building, al…
Security researchers at cybersecurity firm Hacktron AI used Anthropic's Claude Opus 5 to exploit a vulnerability in OpenAI's Discourse-powered community forum, gaining access to an OpenAI employee's ChatGPT account and r…
Researchers submitted SSD-LLaMA to arXiv on 16 Sep 2026, an SSD-native local Mixture-of-Experts inference system that runs trillion-parameter models at over 1 token/s on a single RTX 5090 with no more than 32GB of RAM. A…
Epoch AI maintains a registry of 85 AI benchmarks spanning mathematics, software engineering, agentic workflows, and games, and has flagged a subset as flawed. The registry includes benchmarks such as a long-horizon codi…