A developer testing four Fable 5 AI agents in a Werewolf social deduction game found that the agents produced plausible strategic accusations based on events that never occurred, such as claiming a pl…
OpenAI's GPT-6 Astra, released last week, has prompted an Anthropic user and developer to resubscribe to Codex, calling it a level above Claude models. The user notes that Anthropic's model documentat…
Anthropic has launched a beta feature for the Claude API that allows developers to add or remove tools mid-conversation without invalidating the prompt cache, addressing a long-standing pain point for…
Anthropic released Claude Fable 5.1 and Mythos 5.1, splitting a single underlying architecture into two offerings tailored for different risk profiles. Fable 5.1 is the general-purpose flagship model …
In a hands-on comparison, a developer found that Anthropic's Claude Fable 5.1 and Fable 5 were indistinguishable on real-world tasks, with no noticeable performance difference. The review, published b…
A new essay argues that as AI performance improves and its integration into nuclear affairs deepens, AI will threaten deterrence by instilling overconfidence in decision-makers, especially during cris…
Anthropic revised its enterprise data retention policy for frontier models after customer pushback, replacing a mandatory 30-day retention rule with the Enterprise Frontier Safeguards program that sto…
OpenAI's GPT-6 Astra has won back a prominent AI reviewer, who now uses it as their default model for most tasks, citing its superior engineering, reliability, and coordination capabilities compared t…
OpenAI launched GPT-6 Astra, which it calls 'the world's most intelligent and aligned model,' with state-of-the-art performance on computer use, browsing, software engineering, cybersecurity, science,…
OpenAI's GPT-6 Astra scores 67 on the Artificial Analysis Coding Agent Index, matching Fable 5 at less than half the cost, and scores 61 on the Intelligence Index, equal to GPT-5.6 Sol but with a 2.5x…
OpenAI's GPT-6 Astra achieved state-of-the-art results on the ARC-AGI benchmark, scoring 63% on ARC-AGI-3 and 99% via a new provider adapter harness, surpassing human performance on 96% of ARC-AGI-3 l…
OpenAI has begun rolling out GPT-6 Astra to cybersecurity customers, with co-founder and president Greg Brockman declaring 'we are now in the AGI era' during a September 3 briefing. The model is the f…
A developer used AI assistants to trace his family tree, gathering more than 600 individuals and reaching back 12 generations in some branches within a month. He treated the project like software deve…
Anthropic announced Enterprise Frontier Safeguards (EFS) on September 1, enabling enterprises to keep Claude traffic data in their own cloud infrastructure instead of Anthropic's servers, addressing d…
Abliteration, a Palo Alto startup founded last year, released abliterated-model-large-v2 on Monday, an AI model built on Z.ai's open-weight GLM-5.3 with safeguards removed so it 'doesn't say no' to da…
In a published test of an AI agent's prediction accuracy, the agent's forecast missed because the prediction was disclosed to the subject before the outcome, highlighting a flaw in the experimental de…
Anthropic released Claude Fable 5.1 on September 1, achieving 90% coverage on the ARC-AGI-2 benchmark at 32% lower cost per task than its predecessor, Fable 5, which cost $5.45 per task. The model als…
An analysis of OpenAI's open-source Codex repository shows a step change in development velocity, with commits per month rising from 98 in May 2025 to 893 in August 2026, and authors with 5+ commits g…
Anthropic's Claude Fable 5.1 keeps the same input and output token prices as Fable 5, with the advertised 25% to 45% savings coming solely from a 75% cut in cache read prices, now about 25 cents per m…
Anthropic released Claude Fable (and Mythos) 5.1, which scores 52.6% on the new Terminal-Bench-Science 0.1 benchmark, up from 24.7% for Fable 5, 29.0% for Opus 5, and 22.4% for GPT-5.6 Sol. In a hands…