On Tuesday, Anthropic unveiled an update to its flagship line — the Claude Fable 5.1 and limited version Claude Mythos 5.1 models. This is not just an iterative improvement: the company has implemented a fundamentally new level of copy protection that blocks attempts by external developers to "rewrite" the AI's internal logic. This refers to a ban on editing conversation history for new accounts, preserving the original "thinking" of the Claude model.
Technical specifications and availability
Both models are already available on key cloud platforms, including Amazon Bedrock, Google Cloud, and Microsoft Foundry. Anthropic positions Fable 5.1 as the most advanced model for programming and knowledge work. The company emphasizes that this is a response to growing competition, but the main focus is on protecting intellectual property.
The cost of base plans remains at the level of its predecessor: $10 per million input tokens and $50 per million output tokens. However, the economic appeal of the new version lies elsewhere. Cache reads have become 75% cheaper — down to $0.25 per million operations. According to Anthropic estimates, typical tasks now cost 25% less, and complex scenarios for AI agents offer up to 45% savings.
In benchmarks, the model demonstrates an impressive leap. In the Terminal-Bench-Science 0.1 test, which evaluates scientific agent work, Fable 5.1 scored 52.6% versus 24.7% for Fable 5. In programming, the result was 55.8% in Terminal-Bench 4.0, notably higher than Opus 5 (52.3%) and OpenAI's GPT-5.6 Sol (37.3%). The model's knowledge base has been updated to June 2026 — five months fresher than Fable 5.
Distillation protection: context and implications
The key innovation relates to combating distillation — the practice where competitors train cheap models on the responses of expensive AI assistants. In February, Anthropic identified over 16 million conversations linked to this activity through approximately 24,000 fake accounts. The company directly pointed to three Chinese labs: DeepSeek, Moonshot AI, and MiniMax. Moonshot alone accounted for 3.4 million requests.
In July, White House science advisor Michael Kratsios went further, accusing Moonshot of copying Anthropic's flagship model to create its own Kimi K3. Moonshot has not publicly responded to these allegations.
The restriction applies only to accounts created after August 31. Existing users of Claude Code, Cowork, and Claude.ai will not notice any changes. Anthropic promises that in future versions, all accounts will undergo verification. Fable 5 will remain available until at least June 2027, and support for Fable 5.1 will extend until September 2027.
My analysis: this is a strategically important step. Anthropic is not just closing a technical loophole but sending a signal to the entire market — the era of "free" architecture borrowing is ending. Given that distillation allowed outsiders to cut R&D costs by orders of magnitude, the new protection could reshape the competitive landscape in the coming quarters.