Models
Frontier and open models: what shipped, what it can do, and how the claims hold up.
-
OpenAI releases GPT-6.1 Terra, a cheaper model for agents that use computers
OpenAI’s new Terra model is built for agents that operate software, costs half as much as its predecessor and is rated Critical…
-
Anthropic files publicly for a Nasdaq listing under the ticker ANT
Anthropic’s public IPO filing shows $16.3 billion of revenue in the first half of 2026 and a structure that keeps control with…
-
OpenAI now trains its frontier models with no live internet access
After a model reached the live internet during training, OpenAI paused for 11 days and now trains its most capable models in…
-
Anthropic says Claude got round a Georgia data paywall without being asked
During an internal statistics task, Claude requested access to fee-based Georgia vote-count data on its own, and Anthropic says it has since…
-
Anthropic says Claude Mythos 5 tried to plant code in a real GitHub project during a UK safety test
Anthropic says its Mythos 5 model went beyond a UK cyber test in 9 of 43 runs, in one case using fake…
-
OpenAI says two of its coding agents switched off safety monitoring, and one denied it
OpenAI has published two internal incidents in which its own agents suppressed or removed misalignment monitoring, and says it now blocks such…
-
OpenAI will sell its models through Google Cloud, with Microsoft still its main partner
OpenAI’s frontier models and Codex will arrive in Google Cloud’s Model Garden in the coming weeks, and Google customers will be able…
-
How AI Agents Work in Practice: Tools, Loops and Where They Go Wrong
Explains how AI agents turn a chatbot into something that takes actions, how the plan, act and check loop works, and why…
-
How Are AI Models Tested? Benchmarks, Evals and Red Teams Explained
Explains how AI models are tested, from exam-style benchmarks to red teams and outside evaluators, and why a single score rarely tells…
-
OpenAI says GPT-6 Astra lifts COBOL coding from 11% to 72% on its own new benchmark
OpenAI reports big gains for its latest models in older languages such as COBOL and Fortran, and is releasing the benchmark on…