Claude Fable 5.2 and Mythos 5.2: Pricing, API, Access, Benchmarks

👁 119 reads

Updated: 9 November 2026

Update, 9 Nov 2026 (evening): I have since tested Fable 5.2 against Opus 5.5, GPT-6.1 Sol and Gemini 4 Argon myself - see “My test” below.

Claude Fable 5.2 and Claude Mythos 5.2 came out on 9 November 2026. This page collects everything that matters if you want to use them: what the two models are, what they cost through the API and in the Claude plans, where they are available, how they compare with Claude Opus 5.5, GPT-6 Astra and Gemini 4 Carbon, and what to keep in mind before you switch. Every number here comes from Anthropic or from the other vendors’ own pages; I mark clearly what is a vendor claim. I have not tested Fable 5.2 myself yet. The launch news is here: Anthropic releases Claude Fable 5.2 and Mythos 5.2.

What Claude Fable 5.2 and Mythos 5.2 are

Both are the same model trained by Anthropic, released in two configurations. Fable 5.2 is the general version: everyone with a paid plan or an API account can use it, and it runs with additional classifiers that block high-risk cybersecurity requests (penetration testing, exploits, binary vulnerability analysis) and most research-level biology. When a request is blocked, Fable 5.2 does not just refuse: cyber questions are answered by Opus 4.8 and biology questions by Opus 5. Mythos 5.2 has no such blocks and is only for vetted organizations.

Anthropic positions Fable 5.2 above Opus 5.5 for the hardest technical work: long agentic coding, multi-day agent teams, research with many tools. Opus 5.5 remains the recommended “daily driver”, and the developer docs say to start with Opus 5.5 and move to Fable 5.2 when Opus at a higher effort level is not enough. Both 5.2 models generate output up to 30% faster than the 5.1 generation, according to Anthropic.

How much Claude Fable 5.2 costs

Through the API, Fable 5.2 costs $8 per million input tokens and $40 per million output tokens. Fable 5.1 cost $10 and $50, so the price dropped by 20%. The full 1M-token context is billed at the standard rate, with no long-context surcharge. Cache reads cost $0.20 per million tokens (0.025x the input price), and US-only inference (inference_geo: "us") adds a 1.1x multiplier.

In the Claude apps, Fable is not on the Free plan. Pro costs $20 a month ($17 a month billed annually) and includes Fable only during the launch week until 16 November 2026, capped at 50% of the weekly limit and with one limit reset; after that Pro users pay for Fable with usage credits. Max ($100 a month) includes Fable up to half of the weekly limit, and so does Ultra ($200 a month). Max, Ultra, premium Team seats and premium legacy Enterprise seats also get one to three free resets depending on how long they have been paying (under 6 months: one, 6 to 12 months: two, over 12 months: three), valid until 8 March 2027. Anthropic lists plan prices in dollars, before tax.

Availability

Fable 5.2 launched on all of Anthropic’s platforms at once: the Claude API, Claude Platform on AWS, Google Cloud (both Anthropic-managed and Google-managed) and Microsoft Foundry. On launch evening the models table in the docs showed no separate Amazon Bedrock ID. Fast mode works in Claude Code, on the Claude Platform, on Google Cloud and in Microsoft Foundry (Anthropic hosted). Through third parties, OpenRouter listed anthropic/claude-fable-5.2 at the same $8 / $40 on the evening of 9 November; Mythos 5.2 was not there.

Anthropic’s Supported Regions Policy covers the apps (Claude.ai, Claude Code, Cowork) and the API in the same list of countries, and it does not restrict individual models. The launch announcement does not mention any regional exclusions. By default, Fable comes with 30-day data retention for safety monitoring; eligible customers can get zero data retention.

Fable 5.2 vs Opus 5.5, GPT-6 Astra and Gemini 4 Carbon

The table below comes from Anthropic’s announcement. Fable and Opus were run with safeguards at the highest available effort unless noted; scores for GPT-6 Astra and Gemini 4 Carbon are what Anthropic reports. Treat it as the vendor’s own comparison, not an independent benchmark.

Two things stand out. Fable 5.2 leads in most rows, but Gemini 4 Carbon wins AutomationBench and comes close on Chartography. And Anthropic stresses efficiency more than peak scores: on FrontierCode 2.0, Fable 5.2 at High effort scored 45.3%, roughly Opus 5.5’s best result (45.5%), at about half the cost per task. On Humanity’s Last Exam, Fable 5.2 at Low effort matched Opus 5.5’s best score for less than a tenth of the cost.

Prices from the vendors’ own pages: GPT-6 Astra costs $10 per million input tokens and $50 per million output tokens (up to 272K input tokens), so Fable 5.2 is cheaper on both counts. Gemini 4 Carbon was announced by Google on 28 October 2026 at an introductory $2 and $10, later $4 and $20, with access going first to partners in Google’s Fairwind program.

Mythos 5.2 in cybersecurity and biology

For Mythos 5.2 Anthropic compares results without safeguards. On CyScenarioBench it scored 76.3% against 61.7% for Mythos 5.1, 67.6% for Opus 5.5 and 61% for GPT-6 Cyber. On CVE Rediscovery it reached 89.6%; in an October 2026 run on 1,829 CVEs it found 10.2% more vulnerabilities than Opus 5.5. Mozilla reports roughly twice as good vulnerability-detection metrics as with Mythos 5.1. To get access you apply through the Cyber Verification Program or the Life Sciences Verification Program; Anthropic gives no public price list beyond the API rate of $8 / $40.

Should you switch?

If you use Claude for writing, everyday analysis or standard coding, Opus 5.5 at half the price is still the reasonable default, and that is Anthropic’s own advice. Fable 5.2 makes sense for long agent runs, large migrations, hard research tasks and computer-use automation, where a higher success rate can save more than the price difference. If you used Fable 5.1, moving to 5.2 is an easy call: the price is lower and the vendor scores are higher on every benchmark in the table. My review of the previous version is here: Claude Fable 5.1 pricing, features and review.

My test: Fable 5.2 vs Opus 5.5, GPT-6.1 Sol and Gemini 4 Argon

On launch evening I ran the same 27 tasks through all four models via OpenRouter (medium effort): coding with 136 unit tests, 13 logic puzzles, 10 questions about 3 images and 3 writing tasks. The writing tasks were in Polish, because the test was built for my Polish site. Mythos 5.2 was not available to test.

• All four models scored 100% on code, puzzles and vision, so this set did not separate them.
• Speed: a 500-word essay took Fable 5.2 a median 19.1 s against 26.2 s for Fable 5.1, 27% less time, in line with Anthropic’s “up to 30% faster”.
• Cost of the 27 tasks: Fable 5.2 $0.75, Opus 5.5 $0.52, Gemini 4 Argon $0.14, GPT-6.1 Sol $0.11.
• Fable 5.2 costs twice as much per token as Opus 5.5, but only 1.44 times as much per task, because it used fewer tokens (16,434 vs 23,839).
• Median time to first token: Fable 5.2 2.24 s, Opus 5.5 2.65 s, GPT-6.1 Sol 1.48 s, Gemini 4 Argon 3.44 s.

Full results and raw data (in Polish): Claude Fable 5.2 test on promptowy.com.

More on Claude Fable 5.2 and Mythos 5.2

• Claude Fable 5.2 vs Opus 5.5: price, benchmarks, which to pick
• Claude Fable 5.2 pricing and API: costs, cache, Batch, model ID
• Claude Mythos 5.2: what it is and who can use it
• Claude Fable 5.2 vs GPT-6.1 Sol vs Gemini 4 Argon
• Claude Fable 5.2 and Mythos 5.2 release news

FAQ

When was Claude Fable 5.2 released?

On 9 November 2026, together with Claude Mythos 5.2.

How much does Claude Fable 5.2 cost?

$8 per million input tokens and $40 per million output tokens through the API; $16 and $80 in fast mode; $4 and $20 through the Batch API. In the apps it is included in Max and Ultra up to half the weekly limit, and in Pro only during the launch week.

Can I use Fable 5.2 on the Free plan?

No. Fable is available on Pro, Max, Ultra, Team and Enterprise, and through the API.

How do I get access to Claude Mythos 5.2?

Through Anthropic’s Cyber Verification Program (security professionals) or the Life Sciences Verification Program (researchers). Claude Enterprise customers can use Claude Security, which runs on Mythos 5.2.

Is Fable 5.2 better than Opus 5.5?

According to Anthropic it is a step above Opus 5.5 in almost all evaluated capabilities, for example 69.9% vs 62.3% on FrontierSWE v2. It also costs twice as much per token, and Anthropic still recommends Opus 5.5 as the default.

How large is the context window?

1 million tokens, with up to 128K output tokens. Anthropic gives a reliable knowledge cutoff of June 2026.

Is Fable 5.2 available on Amazon Bedrock?

Anthropic says Fable 5.2 is on AWS, and Claude Platform on AWS has the model ID. On launch evening the docs showed no separate Amazon Bedrock ID.

Kinetyra