Anthropic launches Fable 5.1 and Mythos 5.1 for coding

Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, its most advanced AI models to date, claiming they set a new standard for coding, knowledge work, and long-running problem-solving tasks. The update, announced on September 1, 2026, comes three months after the launch of its predecessor, Fable 5, on June 9, 2026. The company states that Fable 5.1 and Mythos 5.1 are the same underlying model but operate with different levels of safeguards. Fable 5.1 is generally available to all users, while Mythos 5.1 is restricted to vetted professionals in cybersecurity and life sciences through Anthropic's trusted access programs. This includes a new Life Sciences Verification Program for scientific research. Anthropic reports significant performance gains on key benchmarks. Fable 5.1 scored 52.6 on TerminalBenchScience 0.1, more than double the 24.7 score of Fable 5. On TerminalBench 4.0, which tests agentic coding, Fable 5.1 achieved 55.8, up from Fable 5's 42.0. The model also outperformed Anthropic's own Opus 5, released in July, on every published benchmark. A major change is the cost reduction.