Anthropic cuts its top Claude API tier to $8 input and $40 output

Anthropic cuts its top Claude API tier to $8 input and $40 output

9 November 2026: Anthropic has priced Claude Fable 5.2 and Claude Mythos 5.2 at $8 per million input tokens and $40 per million output tokens, 20% below the $10 and $50 still listed for the 5.1 versions. A team on 5.1 pays less by changing its model identifier to claude-fable-5-2, provided its evaluations hold.

Not yet independently verified. Anthropic's announcement says only "a lower price", and the comparison with 5.1 comes from the rates on its pricing page. Google Cloud sets its own prices on its Google-managed route, so a bill there may differ. We will update this when it can be confirmed, and remove this note.

The rate card, line by line

The pricing page shows these per million tokens for Fable 5.2, with the 5.1 rate in brackets:

• Input: $8 ($10)
• Output: $40 ($50)
• Cache write, five minutes: $10 ($12.50)
• Cache write, one hour: $16 ($20)
• Cache hit: $0.20 ($0.25)

Mythos 5.2 carries the same rates as Fable 5.2. Fable 5, the earlier version, remains at $10 and $50 too, with a cache hit price of $1. The new tier still sits well above Claude Opus 5.5, which the page lists at $4 input, and Anthropic itself recommends starting with Opus 5.5 for most coding and knowledge work and moving up only when the job demands it.

Fast mode costs double

An optional fast mode generates tokens up to 2.5 times faster for $16 per million input tokens and $80 per million output tokens. It is available in Claude Code, on the Claude Platform, on Claude Models on Google Cloud and on Microsoft Foundry deployments hosted by Anthropic. For Mythos, fast mode is limited to organisations in Anthropic's Verification Programs.

For interactive tools where a person is waiting, that may be cheap. For a batch job it is simply twice the price.

Fable and Mythos differ in who can use them

The two names cover the same underlying model. Fable has safeguards against misuse in cybersecurity and biology, which is why it is open to all users, while Mythos is reserved for organisations accepted into the Cyber or Life Sciences Verification Programs. Claude Enterprise customers can also use Mythos 5.2 through the Claude Security product.

Availability covers all of its platforms, including Amazon Web Services, Google Cloud and Microsoft Azure.

Anthropic says output comes up to 30% faster than 5.1 even without fast mode. On its own FrontierCode 2.0 benchmark it reports Fable 5.2 at 49.1% at the highest effort setting, against 32.4% for Fable 5.1 and 45.5% for Opus 5.5. Those are the vendor's numbers and methodology.

A data retention condition to read first

Teams on zero data retention should read the safety section before moving. Anthropic says powerful models rely on safeguards that watch usage across multiple requests, and that eligible customers can continue on zero data retention before implementing its Enterprise Frontier Safeguards monitoring. For other customers it states 30-day retention for safety monitoring, so a compliance review belongs ahead of the model swap.

Our AI model category lists 155 models, and both of these are on our launch board.

Sources

• Anthropic, Claude Fable 5.2 and Claude Mythos 5.2, 9 November 2026
• Claude Platform documentation, pricing (read 2026-11-10), 10 November 2026
• InfoWorld, Anthropic Fable 5.2, 10 November 2026