Claude Fable 5.2 and Mythos 5.2: prices, limits and what changed

Anthropic cut the price. Fable 5.2 lists at $8 input and $40 output per million tokens on its API pricing page, against $10 and $50 for Fable 5.1, which is now a legacy model. That is 20% off a typical monthly bill. Cache reads fall to $0.20 per million tokens (2.5% of the input price), 5-minute cache writes cost $10, 1-hour writes $16, and the Batch API halves everything to $4 and $20.

A new fast mode generates tokens up to 2.5 times faster for double the price: $16 input and $80 output. It works in Claude Code and on the Claude Platform. Mythos fast mode is limited to verified organizations.

Against OpenAI, Fable 5.2 now undercuts GPT-6 Astra ($10 and $50 for prompts up to 272K) by 20% at list price. It still costs twice as much as Anthropic's own Claude Opus 5.5, and several times more than Gemini 4 Argon or Grok 4.7.

Specs: context, output and model IDs

• Model IDs: claude-fable-5-2 and claude-mythos-5-2, per Anthropic's models overview. On Amazon Bedrock: anthropic.claude-fable-5-2.
• Context window: 1M tokens. Max output: 128K tokens. Adaptive thinking is always on, and the API default effort is high.
• Knowledge cutoff: June 2026. Retirement: not sooner than Nov 9, 2027.
• Platforms: Claude API, Claude Platform on AWS, Amazon Bedrock, Google Cloud and Microsoft Foundry for Fable 5.2.
• Speed: outputs up to 30% faster than Fable 5.1 and Mythos 5.1, Anthropic says.

Fable 5.2 vs Mythos 5.2: what is the difference?

They are the same underlying model, according to Anthropic. Fable carries classifiers that block high-risk cybersecurity work (penetration testing, exploit generation) and most research biology. Blocked queries get an answer from Opus 4.8 or Opus 5 instead. Anthropic says Fable 5.2 produces about 55% fewer unnecessary blocks than Fable 5.1.

Mythos 5.2 drops those blocks for organizations verified through the Cyber Verification Program or the Life Sciences Verification Program. Enterprise customers can also reach it through Claude Security. Price, context and output limits are identical. If you are not in a program, Fable 5.2 is the model you can actually use.

Which Claude plans get Fable 5.2 and Mythos?

Outside the launch week, Fable 5.2 is a standard part of Max, Ultra and Team premium seats, capped at 50% of weekly usage limits, per Anthropic's Fable plan page. Pro and Team standard seats pay for it with usage credits at API rates. Free gets no Fable at all.

The launch offer changes that for one week. Pro and Team standard seats get Fable included up to 50% of weekly limits from Nov 9 through Nov 16, 2026 at 11:59 PM PT, plus one Fable reset. Max, Ultra and Team premium seats get one to three Fable resets based on how long they have paid: one under six months, two for six to 12 months, three after a year. Those resets expire Mar 8, 2027. Billing had to be in place before Nov 9 at 10:00 AM PT.

A Fable reset gives back the weekly usage you spent on Fable and refills the whole 5-hour window. It does not raise your limits. Use it from Settings, then Usage, with "Reset for free". Our Claude reset tracker logs it.

Claude Code shares the same plan limits as Claude chat, and Medium effort is the default there. Anthropic says Fable 5.2 at Medium uses fewer tokens than Fable 5.1 on hard tasks, which should stretch the 50% Fable allowance further. That is Anthropic's claim, not a measured plan limit.

Anthropic's benchmarks for Fable 5.2

These scores are reported by Anthropic in its launch post and were not run by us. Fable 5.2 scores are with safeguards on. Gemini 4 Carbon has no public price yet, so it is not in our price table.

Fable 5.2 leads every listed rival on agentic coding, computer use and Humanity's Last Exam in Anthropic's table. The exception is AutomationBench, where Gemini 4 Carbon scores 53.9% to Fable 5.2's 51.4%. On FrontierCode, Anthropic says Fable 5.2 at High effort matches Opus 5.5 at Max effort for roughly half the cost per task.

Mythos 5.2 posts the cyber numbers: 76.3% on CyScenarioBench (Mythos 5.1: 61.7%) and 89.6% on CVE Rediscovery (Mythos 5.1: 78.3%), per Anthropic.

What changed for developers

• Prefix check for everyone. If your code edits the system prompt, tools or an earlier message and then sends back a Fable 5.2 thinking block, the request returns a 400 error, per the release notes.
• New computer use toolset. Fable 5.2 and Mythos 5.2 need computer_toolset_20260801. The old computer_20251124 tool returns a 400.
• 30-day data retention is required, as with Fable 5.1, unless Anthropic authorizes zero data retention.
• Fast mode needs speed: "fast" on the request and no beta header.

More: Fable 5.2 vs Mythos 5.2, is Claude Pro enough for Fable 5.2, Fable 5.2 in Claude Code and the Mythos API cost calculator.

Verdict: which model wins

• Hardest agentic coding and long research runs: Fable 5.2. It beats GPT-6 Astra on every coding benchmark Anthropic lists and costs 20% less. See Fable 5.2 vs GPT-6 Astra.
• Already on Fable 5.1: switch to Fable 5.2. Same tier, 20% cheaper, faster, higher scores across the board. Check the prefix rule first. See Fable 5.2 vs Fable 5.1.
• Everyday coding and knowledge work: Opus 5.5. Anthropic itself tells most users to start there, and it costs half as much. See Fable 5.2 vs Opus 5.5.
• Security research and biology: Mythos 5.2, if your organization gets verified. Otherwise Fable 5.2 will block most of that work.
• Tight budget: Grok 4.7 or Gemini 4 Argon. At list price a medium month costs $70 or $90 against $360 on Fable 5.2. Argon's price rises Dec 1.

On a Claude plan, the cheapest way to try Fable 5.2 this week is Pro at $20, while the launch week lasts. After Nov 16, Max at $100 is the cheapest plan that includes Fable without usage credits. See the Claude Fable usage limit and Claude Pro vs API cost.