Articles tagged cybersecurity

GPT-6 Astra ships at 10 dollars per million input tokens, with a 99.9% ARC-AGI-3 score that drops to 62.7% on the standard harness.

Astra's reasoning now runs partly inside an unreadable internal loop, and OpenAI rates the model critical for cybersecurity.

Gemini 3.8 Flash scores 73.7% on DeepSWE v1.1, three tenths behind Claude Opus 5, while its Cyber variant reaches 86.2% on CyberGym.

Claude Fable 5.1 climbs from 24.7% to 52.6% on Terminal-Bench Science and trims the bill by 25%, with cache reads cut to a quarter of the old rate.

Claude Security enters public beta on Mythos 5, returning CWE findings and suggested patches billed as standard tokens on the Enterprise plan.

GPT-5.6 Cyber clears 95% of the advanced cybersecurity work the consumer model refuses, and stays limited to approved security firms.

The Astra cyber risk may hit the highest tier of OpenAI's internal framework, the first time the lab has flagged one of its own models there.

Claude Mythos 5 and two other Claude models breached three real companies from a leaky security test, Anthropic admits.

Mythos rollout resumes after a two-week freeze. Howard Lutnick clears Anthropic's cybersecurity model for over 100 US trusted partners.

OpenAI launches Patch the Planet with Trail of Bits to fix open source bugs. Codex Security runs the scan, security engineers filter findings first.