Skip to content
SI.info

Claude Fable 5.2 and Mythos 5.2: Anthropic Undercuts GPT-6 Astra on Price Days Before Its IPO

Published 3 min read
Anthropic CEO Dario Amodei, in glasses and a blue shirt, gestures with one arm raised while speaking on stage in front of a TechCrunch Disrupt backdrop.
Anthropic CEO Dario Amodei, who called for slowing SI development in September. Fable 5.2 is the company's first frontier release since. Credit: TechCrunch, CC BY 2.0, via Wikimedia Commons
On this page
  1. What launched
  2. Price and specs
  3. What Anthropic claims they can do
  4. “Pacing, not pausing”
  5. The IPO backdrop

Two months after its CEO asked the industry to slow down, Anthropic has released its most capable models yet, and cut the price. Claude Fable 5.2 and Claude Mythos 5.2 launched on Monday, November 9, as the company prepares for a stock market debut that could come as soon as next week.

What launched

  • Claude Fable 5.2 is generally available in the Claude apps and on the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry. Anthropic describes it as built “for demanding reasoning and long-horizon agentic work”. Its model ID is claude-fable-5-2.
  • Claude Mythos 5.2 is restricted. It is offered only to vetted organizations through Anthropic’s Life Sciences and Cyber Verification Programs, and to Claude Enterprise customers through Claude Security, according to CNBC and Fortune.

Mythos and Fable share an underlying model. Anthropic first introduced Mythos in April and limited access because of its advanced cybersecurity capabilities. Fable, a version with extra safeguards, followed about two months later for general use.

Price and specs

Claude Fable 5.2Claude Fable 5.1GPT-6 Astra
Input (per 1M tokens)$8$10$10
Output (per 1M tokens)$40$50$50
Cached input (per 1M tokens)$0.20$0.25$1.00
Context window1M tokens1M tokensAbout 1.05M tokens
Max output128K tokens128K tokens128K tokens

Prices are list prices from Anthropic’s documentation and, for Astra’s cached rate, Fortune. Fable 5.2 also has a June 2026 knowledge cutoff, always-on adaptive thinking, and a retirement date no sooner than November 9, 2027. The cached-input gap is the striking number. SI agents re-send the same long context over and over, so for them Fable 5.2’s cached input is 80% cheaper than Astra’s. That can matter more than the headline rate.

Anthropic says both new models produce outputs about 30% faster than the 5.1 generation. Fable 5.1 is now listed as a legacy model but is still available.

What Anthropic claims they can do

CNBC’s report includes no independent benchmark scores. Anthropic says the models outperform competitors on coding evaluations, and Fortune relayed several of the company’s claims:

  • A team of 1,000 agents running Fable 5.2 formalized a mathematical proof in 30 hours. An earlier unreleased research version of Claude took 11 days.
  • The models designed three working electronic devices from a one-paragraph prompt.
  • An unnamed security company found that one in six confirmed bugs found by Mythos 5.2 were missed by every other frontier model it tested.

Treat these as company claims until outside evaluations appear.

Two men shake hands in a bright office lobby with timber A-frame seating booths and a large indoor tree.
Anthropic's San Francisco office, during a visit by then-UK Technology Secretary Peter Kyle in October 2024. Credit: Alecsandra Dragoi / Department for Science, Innovation and Technology, CC BY 2.0, via Wikimedia Commons

“Pacing, not pausing”

The release is Anthropic’s first frontier launch since CEO Dario Amodei called for slowing development in September. That raises an obvious question, and Anthropic answered it directly. “We called for pacing, not pausing,” Penn, Anthropic’s head of product management for research, told CNBC. The company says the models went through safety and alignment testing, and that it built new evaluations and post-launch monitoring tools for them.

Penn positioned Fable 5.2 for power users with hard engineering, mathematics and robotics problems. For everyday work, Anthropic still points developers to Claude Opus 5.5, at half the price.

The IPO backdrop

The timing is hard to ignore. According to CNBC, Anthropic is preparing for its roadshow and could go public as soon as next week. It is targeting a valuation of $1 trillion to $1.5 trillion, down from a $2 trillion figure floated in recent months. Its amended prospectus reports $33.24 billion in revenue for the nine months to September 30, up roughly 1,200% from a year earlier. Fortune framed the price cut as a direct challenge to OpenAI’s GPT-6 Astra, which launched on September 3 at $10/$50.

Compare: GPT-6 Astra vs Claude Fable 5.2 · Lab: Anthropic · Tracker: All models

Frequently asked questions

How much does Claude Fable 5.2 cost?

$8 per million input tokens and $40 per million output tokens on the Claude API, with prompt-cache reads at 2.5% of the input price ($0.20 per million). Batch requests are 50% off.

What is the difference between Claude Fable 5.2 and Claude Mythos 5.2?

They are built on the same underlying model. Fable 5.2 is generally available with extra safeguards. Mythos 5.2 has some of those safeguards lifted and is available only to vetted organizations through Anthropic's Life Sciences and Cyber Verification Programs, and to Claude Enterprise customers through Claude Security.

Is Claude Fable 5.2 cheaper than GPT-6 Astra?

Yes, by list price. Fable 5.2 costs $8/$40 per million tokens against Astra's $10/$50, about 20% less. For cached input, Fable 5.2 costs $0.20 per million tokens against Astra's $1.00, according to Fortune.

Sources

  1. Anthropic releases Claude Fable 5.2 and Claude Mythos 5.2 ahead of IPO — CNBC, November 9, 2026
  2. Anthropic debuts updated Claude Fable and Mythos models, priced 20% below OpenAI's GPT-6 Astra — Fortune, November 9, 2026
  3. Models overview — Anthropic
  4. Claude Platform release notes — Anthropic

Written by

· Editor

Editor of SI.info. Writes about Super Intelligence, technology policy and the people building frontier models.

How we research and fact-check

Free newsletter

Get The SI Brief

One short email a week: what changed in Super Intelligence, policy and models — and why it matters.

Free. One email a week. Sent via beehiiv, which counts opens and clicks. Unsubscribe anytime. Privacy policy