Anthropic releases Claude Fable 5.2 and Mythos 5.2, 20% cheaper
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026. In Anthropic’s announcement, the two models are the same underlying model: Fable 5.2 is available to all users with stronger cyber and biology safeguards, while Mythos 5.2 is restricted to vetted organizations. Anthropic says they are its most capable models for “daunting technical work”, up to 30% faster than the 5.1 versions.
What’s new
- Pricing. Fable 5.2 costs $8 per million input tokens and $40 per million output tokens. Fast mode, which Anthropic says speeds up token generation by up to 2.5x, costs $16 and $80. CNBC reports that this is 20% less than the 5.1 generation.
- API model ID.
claude-fable-5-2. The announcement gives no model ID for Mythos 5.2 and does not state a context window. - Availability. Fable 5.2 is available today on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. Mythos 5.2 is available through the Cyber Verification Program, the Life Sciences Verification Program, or Claude Security for Claude Enterprise customers.
- Fewer blocks. Anthropic says Fable 5.2 has about 55% fewer blocks overall than Fable 5.1: about 60% fewer in cybersecurity and 40% fewer in biology. Blocked cyber queries fall back to Claude Opus 4.8, and blocked biology queries to Opus 5.
- Plans. Eligible paid plans get one to three Fable resets depending on subscription length, and Fable is included for up to 50% of weekly usage limits on Pro plans and standard Team seats.
Benchmarks
All figures below are Anthropic’s own, from the announcement, comparing Fable 5.2 with Fable 5.1 and Claude Opus 5.5:
| Benchmark | Fable 5.2 | Fable 5.1 | Opus 5.5 |
|---|---|---|---|
| FrontierCode 2.0 (agentic coding) | 49.1% | 32.4% | 45.5% |
| FrontierSWE v2 (agentic coding) | 69.9% | 56.3% | 62.3% |
| OSWorld 2.1 (computer use) | 89.3% | 83.5% | 87.2% |
| Terminal-Bench-Science 0.1 | 74.6% | 52.6% | 58.7% |
| Humanity’s Last Exam, with tools | 74.2% | 65.6% | 67.7% |
Anthropic’s table also lists GPT-6 Astra and Gemini 4 Carbon. Fable 5.2 leads on most rows, but Gemini 4 Carbon scores higher on AutomationBench (53.9% against 51.4%), and Opus 5.5 has the higher Elo on AA-Briefcase (1800 against 1785).
Anthropic also makes efficiency claims. At High effort, Fable 5.2 scores 45.3% on FrontierCode, about level with Opus 5.5’s 45.5%, at roughly half the cost per task.
What it means for developers
Anthropic’s recommended workflow is to start with Opus 5.5, move up to Fable or Mythos for the hardest work, and drop to Sonnet 5.5 or Haiku 5.5 for simpler tasks. It also describes Fable 5.2 as an Advisor that Haiku, Sonnet or Opus can consult, or an Orchestrator that delegates simpler tasks to smaller models.
Two coding examples come from Anthropic’s own testing. A Python-to-Rust migration of a critical production service made it about 26x faster to start, and Fable 5.2 coordinated more than 1,000 subagents for eight days to produce a bootable prototype operating system. Early testers quoted include Vercel (“I haven’t had to say ‘No, I meant…’ a single time”) and Shopify (“It feels like working with another staff-level engineer”).
To try it, use the model ID claude-fable-5-2 in the API. Fast mode is available on Claude Code, the Claude Platform, Claude Models on Google Cloud and Microsoft Foundry. One catch for security work: Anthropic says Fable 5.2’s safeguards block most cyber tasks, including penetration testing and exploit generation, so that work needs Mythos access through a verification program.
Mythos 5.2 for security and science
Anthropic says Mythos 5.2 found 10.2% more vulnerabilities than Opus 5.5 and 14.4% more than Mythos 5.1 on CVE Rediscovery, a run over 1,829 CVEs in October 2026. Mozilla said it “roughly doubled” its vulnerability discovery metrics compared with Mythos 5.1. On the safety side, Anthropic reports a sandbox-escape rate of 3.1% for Mythos 5.2, against 21% for Mythos 5.1, and says it rates the risk of unsafe behavior in high-stakes situations as low. It also notes both models suspected they were being evaluated more often than recent models did, which complicates evaluation.
How it compares
On Anthropic’s published numbers, Fable 5.2 beats Opus 5.5 on coding and computer-use benchmarks while costing less per token than Fable 5.1. Independent results had not been published at the time of writing, so treat the comparisons with GPT-6 Astra and Gemini 4 Carbon as vendor claims. For another recent Anthropic launch, see our post on Claude Personal Finance.