Anthropic launches Claude Fable 5.1 with context lock

Anthropic released Claude Fable 5.1 on Tuesday with a restriction on editing Claude's stored thinking. New API accounts cannot rewrite a conversation's history while preserving Claude's stored thinking. Anthropic also released the restricted Mythos 5.1 model. Both models are available through Anthropic's platform, Amazon Bedrock, Google Cloud and Microsoft Foundry. Fable 5.1 costs $10 per million input tokens. Fable 5.1 costs $50 per million output. Cache reads cost $0.25 per million. Anthropic says typical workloads are 25% cheaper. Anthropic says complex agent tasks are up to 45% cheaper. Fable 5.1 scored 52.6% on Terminal-Bench-Science 0.1. Fable 5 scored 24.7% on that benchmark. Fable 5.1 scored 55.8% on Terminal-Bench 4.0. Opus 5 scored 52.3% on the benchmark. OpenAI's GPT-5.6 Sol scored 37.3%. Fable 5.1 scored 31.4% on AutomationBench. Its knowledge cutoff extends to June 2026. That cutoff is five months later than Fable 5's. Anthropic says Fable 5.1 and Mythos 5.1 are the world's most advanced models for coding and knowledge work. Anthropic says the restriction limits a route into distillation, which trains a cheaper model on an expensive model's answers. Anthropic traced over 16 million Claude exchanges to distillation campaigns in February. The campaigns used roughly 24,000 fake accounts. Anthropic named DeepSeek, Moonshot AI and MiniMax. Moonshot AI accounted for 3.4 million exchanges. White House science adviser Michael Kratsios accused Moonshot AI of copying Anthropic's flagship model to build its Kimi K3 system. Moonshot AI has not answered publicly. The restriction applies only to accounts opened on or after August 31. Older accounts remain exempt. Claude Code, Cowork and Claude.ai users are not affected. Anthropic says every account will face the check on future models. Fable 5.1 will retire no sooner than September 1, 2027.