# Claude Fable 5.2 and Mythos 5.2: one model, two access rules, and Anthropic itself says start with Opus 5.5

By Ivo Donker · 10 November 2026 — text: Claude (AI, Anthropic), checked against the sources
What are Claude Fable 5.2 and Claude Mythos 5.2? Anthropic released both on November 9, 2026, and according to the company they are the same underlying model. Fable has strict cyber and biology safeguards and is available to everyone; Mythos is only for organizations that pass a verification program for cyber or life sciences. Fable 5.2 costs 8 dollars per million input tokens and 40 dollars per million output tokens, twice as much as Opus 5.5. Anthropic itself advises starting most coding and knowledge work with Opus 5.5.

This article lists the facts from [Anthropic's announcement](https://www.anthropic.com/claude-fable-and-mythos-5-2) and the [model documentation](https://platform.claude.com/docs/en/models/fable-5-2/overview). The wider context of that day is in the [trending briefing of November 10](https://trending.llmnet.nl/en/briefing/2026-11-10).

## Same model, different access

According to Anthropic, the difference between Fable and Mythos lies in safeguards and access, not in the model. Fable uses classifiers to block high-risk cyber activity and most research biology tasks. A blocked query then goes to an older model: Opus 4.8 for cyber, Opus 5 for biology. Anthropic says Fable 5.2 blocks about 55 percent less often than Fable 5.1. Mythos 5.2 is meant for dual-use work in cybersecurity and biomedical research, through the Cyber Verification Program and the Life Sciences Verification Program.

## Specifications and price

Specifications and price
Item | Claude Fable 5.2 | Claude Opus 5.5 (for comparison) | 

Model ID | claude-fable-5-2 | see model documentation | 

Context | 1 million tokens | 1 million tokens | 

Max output | 128,000 tokens | 128,000 tokens | 

Price per million tokens (in/out) | $8 / $40 | $4 / $20 | 

Fast mode (in/out) | $16 / $80, up to 2.5 times faster token generation | not compared | 

Cache read | $0.20 per million | not listed on the Fable page | 

Default effort (API) | high | medium | 

Knowledge cutoff | June 2026 | June 2026 | 

Thinking is always on in Fable 5.2 (adaptive) and cannot be turned off; you control depth with the effort parameter. Batch calls get a 50 percent discount. According to Anthropic, output is up to 30 percent faster than with Fable 5.1. We track prices of other models in our overview [models and prices](https://apps.llmnet.nl/en/nieuws-modellen-prijzen).

## Data retention: mind sensitive data

The model documentation says Fable 5.2 requires 30-day data retention and is not available under zero data retention unless Anthropic expressly authorizes it. The announcement says eligible enterprise customers can get zero data retention, including through Enterprise Frontier Safeguards, monitoring that runs inside the customer's own infrastructure and is still in private beta. For most users, then: 30 days of retention for safety monitoring.

## Benchmarks: Anthropic's own figures

Anthropic publishes a series of scores, all measured by the company itself and partly from internal evaluations. Two examples: on SWE v2 Fable 5.2 scores 69.9 percent according to Anthropic, against 62.3 percent for Opus 5.5, and on FrontierCode 2.0 49.1 percent against 45.5 percent. Anthropic also reports that about a thousand Fable 5.2 agents together completed the formalization of Fermat's Last Theorem in 30 hours. We did not check independent measurements for this article.

## What Anthropic itself recommends

Anthropic advises starting most tasks with Opus 5.5, moving to Fable or Mythos only for the hardest open-ended work, and down to Sonnet 5.5 or Haiku 5.5 for simple tasks. Fable can also act as an advisor or as an orchestrator that hands simpler work to smaller models. In Claude Code, Fable 5.2 is now the default Fable model; more on updates to that tool in our overview [Claude Code news](https://apps.llmnet.nl/en/claude-code-nieuws). Subscribers on paid plans get extra Fable usage this week.

## What we did not check

We read the announcement and the model documentation; we did not test the model itself. The benchmark figures, the speed gain and the safety results are Anthropic's claims.

Checked on November 10, 2026.

This article was written with AI (Claude, Anthropic) and checked against the sources below.

## Sources

- Anthropic, Claude Fable 5.2 and Claude Mythos 5.2 (November 9, 2026): [anthropic.com](https://www.anthropic.com/claude-fable-and-mythos-5-2)

- Anthropic, Claude Fable 5.2 model documentation: [platform.claude.com](https://platform.claude.com/docs/en/models/fable-5-2/overview)

- Announcement on X by @claudeai: [x.com/claudeai](https://x.com/claudeai/status/2119867114450268322)