Anthropic
Claude Fable 5.2 Cuts Prices 20% and Tightens a Request Check
Anthropic made Claude Fable 5.2 generally available at $8 and $40 per million tokens, 20% below Fable 5.1, with stricter rules that return errors.

Anthropic has cut the price of Claude Fable by 20% with version 5.2, which is generally available on Amazon, Google and Microsoft clouds. The new version also returns errors on some requests that older accounts could send to Fable 5.1.
Anthropic introduced the model Nov. 9 in a post on its website. The API ID is claude-fable-5-2, and Anthropic built it for “demanding reasoning and long-horizon agentic work.”
Fable 5.2 Costs $8 and $40 per Million Tokens
Anthropic priced Fable 5.2 at $8 per million input tokens and $40 per million output tokens as of Nov. 9, down from $10 and $50 for Fable 5.1. OpenRouter listed the same prices Nov. 12. The model is generally available on the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry.
An optional fast mode runs up to 2.5 times faster at $16 input and $80 output per million tokens.
Mythos 5.2 Is the Same Model With Limited Access
Anthropic offers Mythos 5.2, the same underlying model, only to organizations verified through its Cyber Verification and Life Sciences Verification programs. Fable carries cybersecurity and biology safeguards so it can go to all users.
Anthropic Reports Gains on Coding and Reasoning Tests
Fable 5.2 scored 69.9% on FrontierSWE v2, a test of agentic coding, against 56.3% for Fable 5.1 and 65.5% for GPT-6 Astra. These are Anthropic’s own runs, reported in its Nov. 9 post at the highest available effort with safeguards on. Gemini 4 Carbon led on AutomationBench, 53.9% to 51.4%.
Artificial Analysis ranked Fable 5.2 at Max effort first among 111 models on its Intelligence Index, with a score of 65, as read Nov. 12.
Fable 5.2 Enforces Stricter Request Rules
Fable 5.2 returns a 400 error when code edits the system prompt, tools or an earlier message and then sends back a thinking block from a later turn, according to Anthropic’s Nov. 9 release notes. The check applies to every account. Fable 5.1 applies it by default only to accounts created on or after Aug. 31, 2026.
When a conversation moves from Fable 5.2 to another model, including Opus 5.5 or Fable 5.1, the API drops Fable’s thinking blocks and the request still succeeds. Anthropic recommends starting most coding and knowledge work on Opus 5.5, which it calls its “strongest daily driver.”
Our View
We think the request check matters more than the price for anyone running long agents. A thinking block is the record of the model’s reasoning that the API returns with each reply, and a developer’s code sends it back on the next turn so the model can pick up its own line of thought. If that code has also edited the system prompt, the tools or an earlier message, Fable 5.2 returns a 400 error. Developers who trim tools or rewrite the system prompt mid-run now hit this on every account, including older ones that Fable 5.1 let through.
The 20% price cut is what those developers gain, on both input and output. What they lose is the freedom to change earlier context while carrying thinking blocks forward. They now have to leave earlier turns alone or strip old thinking blocks before they edit.
Routing adds a cost. Anthropic tells developers to start on Opus 5.5, so a run that hands a turn from Fable 5.2 to Opus 5.5 loses Fable’s thinking blocks. The request succeeds, but Opus 5.5 answers without Fable’s reasoning.
The lower price saves money only if the agent code leaves earlier turns untouched, or strips old thinking blocks before it edits them.
