Anthropic Launches Claude Fable 5.1: Can It Stop AI Copycats?
Anthropic launched Claude Fable 5.1 on Tuesday, and the mannequin ships with a lock by itself reasoning. New developer accounts can not rewrite a dialog’s historical past whereas holding Claude’s saved pondering in place.
The firm calls Fable 5.1 and its restricted sibling Mythos 5.1 the world’s most superior fashions for coding and information work. Both went stay throughout Anthropic’s platform, Amazon Bedrock, Google Cloud and Microsoft Foundry.
Follow us on X to get the newest information because it occurs
What Claude Fable 5.1 Costs and Scores
Headline pricing has not moved. Fable 5.1 nonetheless prices $10 per million enter tokens and $50 per million output.
The financial savings sit elsewhere. Cache reads fall 75% to $0.25 per million. Anthropic places typical workloads 25% cheaper, and sophisticated agent duties as much as 45% cheaper.
Anthropic’s personal scorecard exhibits the sharpest achieve in agentic science work. Fable 5.1 hit 52.6% on Terminal-Bench-Science 0.1, greater than double Fable 5’s 24.7%.
The coding lead is narrower. Fable 5.1 took 55.8% on Terminal-Bench 4.0, forward of Opus 5 at 52.3% and OpenAI’s GPT-5.6 Sol at 37.3%.
Business workflow scores practically doubled to 31.4% on AutomationBench. Knowledge now runs to June 2026, 5 months later than Fable 5.
“It’s priced the identical as Fable 5, with 75% cheaper API cache reads. It will get so much additional into a protracted process earlier than it wants your enter, is healthier at telling you when it’s caught, and its writing model is extra pure,” ClaudeDevs noted.
ClaudeDevs is the official updates for builders constructing with Claude AI.
Why the Copying Protection Matters
According to Anthropic, the quieter change targets rivals.
“It is not attainable for brand new API accounts to manually edit Claude’s prior context in a multi-turn dialog whereas preserving the transcript of Claude’s prior pondering,” the launch put up read.
That edit trick is a identified route into distillation, or coaching an inexpensive mannequin on an costly one’s solutions.
Anthropic put laborious numbers on the menace in February. It traced over 16 million Claude exchanges to distillation campaigns run by roughly 24,000 pretend accounts.
The firm named three Chinese labs, DeepSeek, Moonshot AI and MiniMax. Moonshot alone accounted for 3.4 million exchanges.
White House science adviser Michael Kratsios went additional in July. He accused Moonshot of copying Anthropic’s flagship model to construct its Kimi K3 system.
Moonshot has not answered publicly.
The restriction applies solely to accounts opened on or after August 31. Older accounts keep exempt, and Claude Code, Cowork and Claude.ai customers discover nothing.
Anthropic says each account will face the test on future fashions. Fable 5.1 retires no prior to September 1, 2027.
The put up Anthropic Launches Claude Fable 5.1: Can It Stop AI Copycats? appeared first on BeInCrypto.
