Anthropic released Claude Fable 5.2 on 9 November 2026, alongside Claude Mythos 5.2, a version of the same model with different safeguards that is only offered to vetted organizations. Fable 5.2 is the one most developers will use, and the headline is price: it is 20% cheaper per token than Fable 5.1.
The numbers
| Claude Fable 5.2 | Claude Fable 5.1 | |
|---|---|---|
| Input, per 1M tokens | $8 | $10 |
| Output, per 1M tokens | $40 | $50 |
| Context window | 1M tokens | 1M tokens |
| Max output | 128K tokens | 128K tokens |
- Faster: Anthropic says output is up to 30% faster than the 5.1 models.
- Fast mode: an optional mode generates up to 2.5 times faster, at double the price ($16 / $80).
- Where: the Claude API, Amazon Web Services, Google Cloud and Microsoft Azure.
Coding: the biggest change
Anthropic's own figures put Fable 5.2 at 49.1% on FrontierCode 2.0 at its highest effort setting, up from 32.4% for Fable 5.1, and at 69.9 on FrontierSWE v2, against 62.3 for Claude Opus 5.5. Anthropic also says that at its High effort setting, Fable 5.2 roughly matches Opus 5.5 at maximum effort for about half the cost per task.
Independent results are mostly positive. Artificial Analysis gives Fable 5.2 a score of 65 on its Intelligence Index, ahead of Google's Gemini 4 Carbon, though it trails in agentic knowledge work and terminal-based coding.
The catch: cheaper tokens, not always cheaper tasks
A lower price per token doesn't guarantee a lower bill. R&D World reports that at full effort, Fable 5.2 can spend enough extra tokens that the cost per task doubles compared with Fable 5.1, even after the price cut. At lower effort settings it is cheaper. If you move to 5.2, set the effort level deliberately and watch the cost per request, not the per-token price.
Fewer unnecessary refusals
In early testing, Fable 5.2 blocked about 55% fewer harmless requests than Fable 5.1, with the biggest drops in cybersecurity and biology questions, according to launch coverage. Anthropic also says it deliberately limited a few capabilities during training, including accelerator kernel optimization.
On SparrowCodes
Claude Fable 5.1 is available on SparrowCodes today, along with Claude Opus 5.5. For everyday coding where Fable is more than you need, Sparrow Auto picks a lower-cost model per task.