Claude Fable 5.2: New Features, Pricing and API

Claude Fable 5.2 launched on November 9, 2026, with standard API prices of $8 per million input tokens and $40 per million output tokens, according to Anthropic's launch post. That is 20% below Claude Fable 5.1. Anthropic also offers a fast mode at $16 input and $80 output per million tokens.

• Release: November 9, 2026; API ID claude-fable-5-2.
• What changed: standard rates are 20% lower than Fable 5.1 at $8 input and $40 output per million tokens; optional fast mode costs $16 and $80.
• Current specs: 1M-token context, 128k maximum output, always-on adaptive thinking.

The same launch introduces Claude Mythos 5.2, which shares the underlying model with different safeguards.

What's new in Claude Fable 5.2?

Anthropic's what's-new page lists a 1M-token context window as both the default and maximum, plus 128k maximum output tokens. The model uses adaptive thinking that is always on, and the models overview lists high as the default effort level. A 1M window is useful when one task needs a large repository, a long transcript or several logs at once; the 128k output ceiling helps with long reports and multi-file patches.

Anthropic reports improvements in coding, reasoning and vision, and says pre-release testers found the model communicates more clearly. The launch post has the vendor's examples.

Anthropic says outputs are up to 30% faster than Fable 5.1 and fast mode reaches up to 2.5 times the token-generation rate. Standard input and output cost 20% less than Fable 5.1, so the same token counts produce a lower bill before any speed gain.

Fast mode raises token-generation speed at higher token prices. It is available in Claude Code, the Claude Platform, Claude Models on Google Cloud (Anthropic managed and Google managed) and Microsoft Foundry (Anthropic hosted). Test it when output generation dominates waiting time. Generating more tokens per second does not guarantee a shorter total response time.

Claude Fable 5.2 pricing

All rates below are USD per million tokens (MTok) from Anthropic's pricing page and what's-new page. These are direct Claude API standard global rates; AWS, Google Cloud and Microsoft Foundry set their own regional prices. Rates checked November 9, 2026.

The table covers input and output only. Anthropic lists 5-minute cache writes at $10, 1-hour cache writes at $16, and batch at $4 input and $20 output. Standard cache reads are $0.20 on Fable 5.2 and $0.25 on Fable 5.1 per million tokens.

For a request with 100,000 uncached input tokens and 10,000 billed output tokens (including thinking when billed), Fable 5.2 costs $1.20: 0.1 × $8 plus 0.01 × $40. Fable 5.1 costs $1.50 on the same counts. Fast mode on Fable 5.2 costs $2.40. These unit-price calculations exclude cache, tools, taxes and retries.

Where Claude Fable 5.2 is available

Anthropic says Fable 5.2 is available today on all platforms, including AWS, Google Cloud and Microsoft Azure. The IDs differ by platform:

• Claude API: claude-fable-5-2
• Amazon Bedrock: anthropic.claude-fable-5-2
• Claude Platform on AWS, Google Cloud and Microsoft Foundry: claude-fable-5-2
• Fast mode on Microsoft Foundry: claude-fable-5-2-fast

Fast mode has a narrower list than standard mode: Claude Code, the Claude Platform, Claude Models on Google Cloud (both managed options) and Microsoft Foundry (Anthropic hosted). Use the exact ID in configuration and record it with each test run so a later alias change does not blur the result.

Four checks before migrating to Fable 5.2

Claude Code, claude.ai, Claude Managed Agents and Claude Agent SDK users do not need request changes, according to the what's-new page. Custom clients should check four things before switching from claude-fable-5-1.

1. Keep the conversation prefix stable. If a client edits the system prompt, the tool list or an earlier message and then reuses later thinking blocks, Anthropic returns HTTP 400 by default when prefix_mismatch_behavior is unset. Read the migration notes on the what's-new page before changing history; when you need a different prefix, start a fresh conversation.
2. Leave adaptive thinking on. Fable 5.2 always uses adaptive thinking. Keep the thinking field adaptive, then tune effort for the task. The default is high; test a lower effort against it before assuming the default is the cheapest setting.
3. Update the computer-use toolset. The release notes say computer_toolset_20260801 replaces computer_20251124 on Fable 5.2, except on Bedrock. Change the tool definition before testing a computer-use agent, and keep the Bedrock exception if the same code runs on several platforms.
4. Check data retention. Fable 5.2 requires 30-day retention, and zero data retention needs Anthropic's explicit permission. Confirm the policy before sending source code, incident logs or personal data to production.

Fable 5.2 vs Opus 5.5: which should you use?

Anthropic recommends starting with Claude Opus 5.5 for most coding and knowledge tasks. The same models overview says to use Fable 5.2 for demanding reasoning and long-horizon agentic work, or when Opus 5.5 at higher effort still falls short.

The price gap is large: Opus 5.5 is listed at $4 input and $20 output per million tokens, half Fable 5.2's standard rates. Fable 5.2 has to save enough retries, review time or failed runs to justify the difference. Test both on the same repository, prompt, tools and effort setting, then divide total cost by the number of accepted tasks.

Our Opus 5.5 vs Sonnet 5.5 guide covers the cheaper Claude options. The Fable 5.1 review covers the previous Fable generation.

Fable 5.5, Mythos 5.2 and the name confusion

The launch has two names. Fable 5.2 is the generally available model. Claude Mythos 5.2 uses the same underlying model with different safeguards and is limited to vetted organizations through Anthropic's Cyber Verification Program and Life Sciences Verification Program. Mythos results come from a restricted access tier, so check which model and safeguards produced them.

Fable 5.5 is a separate question. The current lineup lists Fable 5.2, Opus 5.5, Sonnet 5.5 and Haiku 5.5; the 5.5 suffix belongs to the Opus, Sonnet and Haiku families, and that lineup does not include a Fable 5.5. Our Fable 5.5 release-status guide tracks any official change.

Which Claude models are on OmniaKey?

OmniaKey's catalog does not include Fable 5.2 as of November 9, 2026. For current Claude access, start with the Claude Fable 5.1 model page or the Claude Opus 5.5 model page, and check the Claude model catalog for live rates and availability. One OmniaKey key works across the models in the catalog.

If Fable 5.2 appears in the catalog later, verify the model ID, price and data terms on its model page before sending production traffic.

Frequently asked questions

What is Claude Fable 5.2?

Claude Fable 5.2 is Anthropic's Fable model launched on November 9, 2026. It has a 1M-token context window, 128k maximum output, always-on adaptive thinking and standard API prices of $8 input and $40 output per million tokens.

How much does Claude Fable 5.2 cost?

Anthropic lists $8 per million input tokens and $40 per million output tokens. Fast mode is $16 input and $80 output. A request with 100,000 input tokens and 10,000 billed output tokens costs $1.20 at the standard rates before cache, tools, taxes and retries.

Is Claude Fable 5.2 cheaper than Fable 5.1?

Yes on standard token rates: input falls from $10 to $8 and output from $50 to $40, a 20% cut. Cache reads fall from $0.25 to $0.20 per million tokens. Total task cost also depends on retries, tools and how much output the task needs.

What is the difference between Fable 5.2 and Mythos 5.2?

They share the same underlying model but use different safeguards. Fable 5.2 is generally available; Mythos 5.2 is limited to vetted organizations through Anthropic's Cyber Verification Program and Life Sciences Verification Program.

Sources checked

• Anthropic: Introducing Claude Fable 5.2 and Claude Mythos 5.2
• Anthropic: What's new in Claude Fable 5.2
• Anthropic: Models overview
• Anthropic: API release notes
• Anthropic API pricing