Models
Claude Fable 5.2 and Mythos 5.2: 20% cheaper at $8/$40, 49.1% on FrontierCode 2.0, already in GitHub Copilot and Claude Code
Anthropic released Claude Fable 5.2 and Mythos 5.2 on November 9 at $8/$40 per million tokens (down from $10/$50), with a 1M-token context and 49.1% on FrontierCode 2.0. The model landed the same day in GitHub Copilot and in Claude Code 2.1.321, where it is now the default Fable model.
Anthropic published Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026 (announcement). Fable and Mythos are the same underlying model: Fable ships with cybersecurity and biology safeguards and is open to all customers, while Mythos is reserved for organizations accepted into Anthropic's Cyber or Life Sciences Verification Programs. The same day, GitHub made Fable 5.2 available in Copilot and Anthropic shipped it in Claude Code.
Price, context and availability
| Item | Claude Fable 5.2 | Claude Fable 5.1 |
|---|---|---|
| API price (input / output, per million tokens) | $8 / $40 | $10 / $50 |
| Fast mode (input / output) | $16 / $80, up to 2.5x faster | — |
| Context window / max output | 1M / 128K tokens | 1M tokens |
| Cache read | $0.20 per million tokens | — |
| Default effort on the API | high | — |
Figures from the Fable 5.2 model page and the existing Fable 5.1 row of our model comparison. Anthropic also says the 5.2 models generate output up to 30% faster than the 5.1 versions.
- Model ID:
claude-fable-5-2on the Claude API, Amazon Bedrock (anthropic.claude-fable-5-2), Claude Platform on AWS, Google Cloud (Anthropic-managed and Google-managed) and Microsoft Foundry. - Data retention: Fable 5.2 requires 30-day data retention for safety monitoring and is not available under zero data retention unless Anthropic expressly authorizes it. Anthropic's Enterprise Frontier Safeguards, which run that monitoring in the customer's own infrastructure, are in private beta (Bridgewater, Salesforce and Stripe are named).
- API changes: adaptive thinking is always on; forced tool use (
tool_choiceanyortool) returns a 400 error, as dotemperature,top_pandtop_kvalues listed in the docs. - Subscriptions: this week, eligible paid plans get one to three Fable resets, and Fable counts for up to 50% of weekly usage limits on Pro and standard Team seats.
Benchmarks published by Anthropic
These are Anthropic's own figures, with safeguards on and at each model's highest available effort unless noted; harnesses differ between labs, so treat cross-vendor comparisons with care.
| Benchmark | Fable 5.2 | Fable 5.1 | Opus 5.5 | GPT-6 Astra | Gemini 4 Carbon |
|---|---|---|---|---|---|
| FrontierCode 2.0 (agentic coding) | 49.1% (Xhigh) | 32.4% | 45.5% | 42.8% | 41.1% |
| FrontierSWE v2 | 69.9% | 56.3% | 62.3% | 65.5% | 61.2% |
| Terminal-Bench-Science 0.1 | 74.6% | 52.6% | 58.7% | 68.1% | 64.8% |
| OSWorld 2.1 (computer use, partial offline set) | 89.3% | 83.5% | 87.2% | 79.5% | 81.5% |
| Humanity's Last Exam (with tools) | 74.2% | 65.6% | 67.7% | 57.2% | — |
| AutomationBench (business workflows) | 51.4% | 31.4% | 42.5% | 41.4% | 53.9% |
On cost, Anthropic says Fable 5.2 at High effort scores 45.3% on FrontierCode 2.0, about level with Opus 5.5's best (45.5%) for roughly half the cost per task. Anthropic still recommends starting most coding work on Claude Opus 5.5 ($4/$20) and moving up to Fable only when evals require it.
Other claims in the announcement, all reported by Anthropic: more than 90% of the changes merged into Anthropic's core codebase are now written by Claude (with human guidance); a team of about 1,000 Fable 5.2 agents formalized Fermat's Last Theorem in Lean in 30 hours; an earlier version of the model migrated a production service from Python to Rust; and Fable 5.2 triggered about 55% fewer safeguard blocks than Fable 5.1 in early testing (60% fewer in cybersecurity, 40% fewer in biology). Blocked queries are answered by Opus 4.8 (cybersecurity) or Opus 5 (biology).
GitHub Copilot: available, off by default for organizations
GitHub announced on November 9 that Fable 5.2 is generally available in Copilot for Pro+, Max, Business and Enterprise users (GitHub changelog). It appears in the model picker of VS Code, Visual Studio, Copilot CLI, the Copilot coding agent, the Copilot app, github.com, GitHub Mobile, JetBrains IDEs, Xcode and Eclipse, with a gradual rollout. It is billed at provider list price under usage-based billing: $8 input, $0.20 cached input, $10 cache write and $40 output per million tokens (Copilot models and pricing).
Business and Enterprise admins must turn on a dedicated policy, off by default, which also acknowledges that Anthropic retains prompts and outputs to run its safety classifiers (GitHub says this data is not used for training). Eligible enterprises can use Fable 5, 5.1 and 5.2 with zero data retention until the end of the calendar year under a temporary exemption; after that, Enterprise Frontier Safeguards will be required.
Claude Code 2.1.321 and 2.1.322
- v2.1.321 (November 9, 19:22 UTC, 20:22 Paris time) adds Claude Fable 5.2 (
claude-fable-5-2) as the default Fable model, adds--settingstoclaude remote-control, and fixes several security issues, including MCP OAuth requests that could send refresh tokens to another host after a redirect and a case where a read-only Bash command ran outside the sandbox without a permission prompt (release notes). - v2.1.322 (November 10, 02:29 UTC, 03:29 Paris time) adds fast mode on Vertex in the global, us and eu regions, falling back to standard speed if Vertex refuses it (release notes).
What to keep in mind
- All benchmark and productivity figures above come from Anthropic; none have been reproduced independently yet.
- The price cut is real on paper, but Fable 5.2 still costs twice Opus 5.5 per token, and its default API effort is
high. - For organizations, the 30-day retention requirement is the main change to check before turning the model on, in the Claude API as in Copilot.
Sources
- Anthropic, "Claude Fable 5.2 and Mythos 5.2", November 9, 2026, checked November 10, 2026
- Anthropic docs, Claude Fable 5.2 model page, checked November 10, 2026
- Anthropic docs, What's new in Claude Fable 5.2, checked November 10, 2026
- GitHub Changelog, "Claude Fable 5.2 in GitHub Copilot", November 9, 2026, checked November 10, 2026
- GitHub Docs, Models and pricing for GitHub Copilot, checked November 10, 2026
- Anthropic, Claude Code v2.1.321 release notes, November 9, 2026, checked November 10, 2026
- Anthropic, Claude Code v2.1.322 release notes, November 10, 2026, checked November 10, 2026
Spotted an error in this article? Report it · Our corrections policy