Claude Mythos

Claude Mythos — explained in more detail

Claude Mythos (internally codenamed “Capybara”) was announced as Mythos Preview on April 7, 2026. The model defines a new tier above the Opus line — larger and more capable than Claude Opus 4.7. Mythos first surfaced publicly on March 26, 2026, when a CMS misconfiguration briefly exposed a pre-release blog draft at Anthropic.

Key features

• Cybersecurity strength: autonomous zero-day discovery across all major operating systems and browsers, autonomous exploit development, vulnerability chaining, reverse engineering of closed-source binaries.
• Benchmarks: SWE-bench 93.9%, USAMO 97.6% — according to Anthropic “far ahead of any other AI model.”
• Jump over Opus 4.6: dramatic improvements in exploit success rates on OSS-Fuzz repositories and Firefox exploit development.

Availability (Project Glasswing)

Mythos is not being made generally available. Anthropic distributes the model via the “Glasswing” program specifically to cybersecurity defenders, critical infrastructure operators, maintainers of important open-source projects, and select industry partners. The goal: defenders should be able to close vulnerabilities before equivalent capabilities become widely available.

Discover more

Blog

AI Recap July 2026: Two Anthropic Models, a Government Review and the Open-Weight Pressure

July 2026 in AI: Sonnet 5 and Opus 5 from Anthropic, GPT-5.6 after a US review, Kimi K3 as open weights, and new GEO controls in Search Console.

Glossary

Claude Fable 5.1

Anthropic's flagship Claude Fable model, released on 1 September 2026. A frontier model for demanding reasoning, coding, research and long-running agents; 1M-token context, multimodal input, proprietary access via API and the Claude apps.

Encyclopedia

Comparing AI models — who builds what and how to choose

The major model families in 2026 at a glance. Who builds Claude, GPT, Gemini, Llama, Mistral, DeepSeek, Qwen — and which model to pick when.

Topic overview

AI ModelsClaude