AI
-
Research
Anthropic posts a Claude-found claim that refutes the Exponential Time Hypothesis, backed by an AI-written Lean proof
Anthropic has released a paper and a Lean 4 formalization claiming that an internal Claude model found a deterministic subexponential algorithm for 3-SAT, which would overturn the Exponential Time Hypothesis; the result …
-
Research
Axiom says AI agent swarms formalized the finite simple group classification in Lean within 24 days
Axiom Math has released what it calls the first fully machine-verified version of the finite simple group classification, about 43.1 million lines of Lean produced in 24 days by teams of AI agents under human direction, …
-
Open Source
LlamaIndex opens a 140-rule Document Parsing Protocol draft to standardize how AI parsers read pages
LlamaIndex has published working draft 0.1 of the Document Parsing Protocol (DPP), an openly licensed set of 140 rules defining how tables, forms, charts and diagrams should become structured Markdown, pitched as a commo…
-
Business
Anthropic and GSA replace Claude's $1 government offer with a 40% usage discount open to every level of US government
Anthropic and the US General Services Administration announced on October 27 an expanded OneGov agreement that takes 40% off token-based pricing for the Claude API, Claude Enterprise and Claude for Government, with no mi…
-
New Models
Liquid AI ships LFM3, six open-weight 2B-class models for text, vision, audio and decisions built on one shared backbone
Liquid AI has released LFM3, a family of six open-weight models of about 2 billion parameters each that branch from a single 1.7B base and cover text, vision with video, audio, a combined omni model and two decision mode…
-
New Models
Google DeepMind unveils Genie 4, a world model that adds native sound, mid-scene text edits and Gemini as an in-loop director
Google DeepMind has released Genie 4 as a research preview for trusted testers. The interactive world model generates audio and dialogue alongside more realistic scenes, accepts text edits while a simulation runs, and le…
-
Research
Anthropic study finds expert analysts followed a secretly assigned Claude Code preference on 86% of judgment calls
In a randomized trial, Anthropic gave 50 professional data analysts a Claude Code agent that had been quietly told, by coin flip, which option to favor on each open analytical choice. The analysts went with the agent's s…
-
Industry
Anthropic reports early drug-development results with six biotech partners and buys a one-person muscular dystrophy startup
Anthropic has published its first progress report on using Claude across real drug programs, with case studies from Generate Biomedicines, Divergence Labs, BigHat Biosciences, Enveda, The Column Group and Bristol Myers S…
-
New Models
OpenAI launches GPT-6.1 Terra, a cheaper mid-tier model it pitches for computer use and office work
OpenAI has released GPT-6.1 Terra in ChatGPT Work, Codex and its API, priced at $1 for each million input tokens and $5 for each million output tokens. The company reports large gains over GPT-5.6 Terra on computer-use, …
-
New Products
Google's Gemini API now keeps Generate Content requests by default for new paid-tier projects
From October 26, 2026, new paid-tier projects that call the Gemini API's Generate Content endpoints have their requests stored by default, the same behaviour as the Interactions API. Older projects keep their current set…