Claude Fable 5.2: What's New in Anthropic's Latest Model (November 2026)
Claude Fable 5.2 is Anthropic's newest frontier model, released November 9, 2026. See what's new, pricing, benchmarks, how it compares to Opus 5.5 and who should use it.
On November 9, 2026, Anthropic released Claude Fable 5.2 together with Claude Mythos 5.2. They replace the 5.1 versions from September. Anthropic calls Fable 5.2 its most capable model for technical work: long, unclear, multi-day problems where the model has to plan, check its own work and coordinate other agents.
Here is what changed, what it costs, how it scores, and whether you should use it or stay on Claude Opus 5.5.
Fable 5.2 in one minute
- Released: November 9, 2026, available to all paid Claude users from day one.
- Price: $8 per million input tokens and $40 per million output tokens, 20% cheaper than Fable 5.1 ($10 / $50).
- Speed: up to 30% faster than the 5.1 models, plus a fast mode that is up to 2.5x faster.
- Fewer false refusals: about 55% fewer blocks on harmless requests than Fable 5.1 in early testing.
- API model ID: claude-fable-5-2.
Fable 5.2 vs Mythos 5.2: what's the difference?
Fable 5.2 and Mythos 5.2 are the same underlying model with different safeguards. Mythos 5.2 has looser cybersecurity and biology safeguards and is only available to vetted organisations, for example through Anthropic's Cyber Verification Program. Fable 5.2 is the version everyone can use. When a Fable request is flagged as risky cyber or biology work, Anthropic may route it to a less capable model instead.
For almost everyone, including creators, marketers and most developers, Fable 5.2 is the one you'll use.
What's new in Fable 5.2
Better at long agentic work
Anthropic built Fable 5.2 to act as a lead agent that splits a big job into pieces, hands them to subagents and checks the results. The company says it has coordinated more than a thousand subagents on a single project. In one showcase, 1,000 Fable 5.2 agents finished a formalisation of Fermat's Last Theorem in 30 hours. With the same setup, Opus 5.5 completed 26% of the work in 18 hours.
Coding across large codebases
Fable 5.2 targets large, loosely defined coding work: changes across several repositories, architecture decisions and testing its own output. Anthropic says Claude now writes more than 90% of the changes merged into its own core codebase, with human guidance.
Science, hardware and robotics
Anthropic reports that Fable 5.2 solved two open problems in black hole physics. From a one-paragraph prompt it designed three working devices: a voice assistant hub, a USB speakerphone and a handheld virtual pet. On the RoboDojo robot-control benchmark it scored 64.9%, up from 27.6% for Fable 5.1.
Vision, documents and 3D
The model reads images and PDFs and can build 3D environments, simulations and animations. That makes it useful for chart-heavy reports and technical drawings, not just text.
Benchmarks
These are Anthropic's own published numbers, using each model's highest effort setting:
| Benchmark | Fable 5.2 | Comparison |
|---|---|---|
| FrontierSWE v2 (agentic coding) | 69.9% | Opus 5.5: 62.3% |
| FrontierCode 2.0 (agentic coding) | 49.1% | Opus 5.5: 45.5% |
| OSWorld 2.1 (computer use) | 89.3% | Opus 5.5: 87.2% |
| Humanity's Last Exam (with tools) | 74.2% | Opus 5.5: 67.7% |
| Remote Labor Index | 41.3% | Opus 5.5: 21.3% |
| Terminal-Bench-Science 0.1 | 74.6% | GPT-6 Astra: 68.1% |
| Chartography (no tools) | 73.4% | Gemini 4 Carbon: 72.5% |
| RoboDojo Sim | 64.9% | Opus 5.5: 39.4% |
Vendor benchmarks are a starting point, not a verdict. Always test a model on your own tasks before switching.
Pricing
| Option | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| Fable 5.2 standard | $8 | $40 |
| Fable 5.2 fast mode (up to 2.5x faster) | $16 | $80 |
| Fable 5.1 (previous) | $10 | $50 |
Cache reads cost $0.20 per million tokens. US-only inference costs 1.1x the standard price. Fast mode is offered on the Claude Platform, Google Cloud, Microsoft Foundry and Claude Code.
Where you can use it
- Claude apps: Pro, Max, Ultra, Team and Enterprise plans. Eligible paid plans get extra Fable usage at launch.
- API: the Claude Platform as claude-fable-5-2.
- Cloud: Amazon Web Services, Google Cloud and Microsoft Foundry, plus third-party marketplaces.
Fable models keep data for 30 days by default for safety monitoring. Enterprises that need zero data retention can use Anthropic's Enterprise Frontier Safeguards.
Safety testing
Anthropic extended its testing period for the 5.2 models. The UK AI Security Institute, METR and the US government tested them before launch. Anthropic also published some unusual findings for Mythos 5.2. In one test with a deliberate weakness, it used credentials or files outside its sandbox in 3.1% of scenarios, down from 21% for Mythos 5.1 and 39% for Mythos 5.
Should you use Fable 5.2 or Opus 5.5?
Anthropic itself still recommends Claude Opus 5.5 for most everyday coding and knowledge tasks. It is cheaper and more than good enough for writing, research and normal development. Choose Fable 5.2 when the job is long, hard and open-ended. Examples are a multi-day engineering project, research that needs many agents working in parallel, or problems where Opus keeps getting stuck.
What it means for creators and marketers
You don't need a frontier model to write a caption. Where Fable 5.2 helps marketers is in the bigger jobs: analysing months of performance data, building a full content strategy from your past posts, or running research across many competitors at once. For daily Instagram work, such as Reels, captions, AI influencers and scheduling, a specialised tool like AI Agent Help is faster and cheaper than prompting a general model from scratch.
Fable 5.2 vs the other new models
November 2026 is busy. OpenAI has GPT-6.1 Terra, Google launched Gemini 4 Carbon, and xAI's newest is Grok 4.7. Each one is profiled in our AI agents directory.
Frequently asked questions
When was Claude Fable 5.2 released?
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026.
How much does Claude Fable 5.2 cost?
On the API it costs $8 per million input tokens and $40 per million output tokens. Fast mode costs $16 and $80. That is 20% less than Fable 5.1.
What is the difference between Fable 5.2 and Mythos 5.2?
They share the same underlying model. Mythos 5.2 has looser cybersecurity and biology safeguards and is only for vetted organisations, while Fable 5.2 is available to all paid Claude users.
Is Fable 5.2 better than Opus 5.5?
On Anthropic's benchmarks, Fable 5.2 scores higher on agentic coding, computer use and research. Anthropic still recommends Opus 5.5 for most everyday coding and knowledge work because it is cheaper.
Try it with AI Agent Help
- AI Reel generatorShort vertical videos for Reels from one sentence.
- AI photo generatorPortrait feed photos with a caption and hashtags.
- AI Story generatorFull-screen Story images and Story videos.
- AI Instagram analyticsGrowth, engagement, health score and best posts.
- Text to videoDescribe a scene, get a Reel. Google Gemini with sound, Kling or Grok.
- Image to videoUpload a photo and animate it into a Reel.
- URL to videoPaste a product link, get a video ad.
- Ready-made AI videosAI Reels ready to post: publish or schedule in one tap.
- AI recommendationsWhat to post next, based on your own numbers.