Analysis
Anthropic released two new flagship models, Claude Fable 5.2 and Mythos 5.2, on Monday, Nov. 9, 2026, pricing them 20% below OpenAI's GPT-6 Astra on ordinary tasks and up to 80% cheaper on cached tokens -- what Fortune called the AI safety company's answer to the GPT-6 Astra model OpenAI shipped in early September, Fortune reported.
What Anthropic released
Fable 5.2 is available broadly across Amazon Web Services, Google Cloud and Microsoft Azure, while Mythos 5.2 access is limited to vetted cybersecurity professionals and life-sciences researchers, with Claude Enterprise customers reaching it through Anthropic's Claude Security product. Anthropic says the models designed three working electronic devices from scratch, solved two open problems in black-hole physics, and had 1,000 agents running Fable 5.2 formalize Fermat's Last Theorem in 30 hours, versus 11 days for an earlier unreleased Claude research version. A separate security-testing firm found that one in six confirmed bugs Mythos 5.2 found were missed by every other frontier model it was compared against. Anthropic says more than 90% of its own code is now written by AI, and recommends the pricier Opus 5.5 -- not these new models -- for everyday tasks.
Pricing: how Fable and Mythos undercut GPT-6 Astra
Fable 5.2 and Mythos 5.2 cost $8 per million input tokens, versus $10 for GPT-6 Astra.
“A separate security-testing firm found that one in six confirmed bugs Mythos 5.2 found were missed by every other frontier model it was compared against.”
Output tokens run $40 per million, versus $50 for GPT-6 Astra -- together a 20% discount on ordinary usage.
The gap widens most on cached tokens, which cost just $0.20 per million versus $1 for GPT-6 Astra, an 80% cut that matters most for prompts carrying large documents or long conversation histories. Anthropic did not disclose a new valuation alongside the release.
Anthropic's funding and IPO backdrop
Pulse previously reported on Anthropic's $3.4 billion deal for 1.7GW of gas generators to fuel its compute buildout; Monday's model launch shows the company applying similar competitive pressure on the product side days ahead of a mega IPO. Fortune frames Fable 5.2 and Mythos 5.2 as potentially Anthropic's last major releases before that listing, timed to re-accelerate revenue growth after Anthropic's own S-1 filing disclosed a third-quarter slowdown.
Anthropic closed a $65 billion Series H at a $965 billion valuation in May 2026, led by Altimeter Capital, Dragoneer Investment Group, Greenoaks and Sequoia Capital, and has since taken on roughly $78 billion in debt financing in 2026 as it builds out compute capacity ahead of the listing.
The AI model market and competitors
The price cuts put Anthropic in direct competition with OpenAI's GPT-6 Astra on cost per token, a front OpenAI has fought hard on since closing an $11 billion funding tranche at a $1.25 trillion valuation in early November. Pricing frontier models this aggressively is a bet that usage volume, not per-token margin, drives revenue at scale -- the same logic Anthropic is testing against a revenue base it has described as running above a $65 billion annualized rate.
The counterweight
However, a 20-80% price cut on new models also compresses gross margin at the exact moment Anthropic's own IPO filing flagged slowing revenue growth, and neither Anthropic nor Fortune disclosed how the new pricing affects per-token margins or whether it signals compute costs falling faster than usage is growing. Impressive capability demos -- a math proof, bug-finding statistics -- are not the same as disclosed revenue, and Anthropic gave no updated user or revenue figures alongside this release.
What to watch
Whether Anthropic's next S-1 amendment shows the price cuts reaccelerating revenue growth, or compressing margins further, will matter more to IPO pricing than the models' benchmark wins.