www.anthropic.comNo verification date supplied. Check the official page.
About this model
Anthropic's AI assistant and model family for writing, analysis and coding.
Official access and developer resources are listed below. Consult the provider for available models, regions and access requirements.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026. Fable 5.2 is generally available as claude-fable-5-2. Access to Mythos 5.2 (claude-mythos-5-2) is limited to approved organizations.
Plans and other pricing
Claude API: Opus 5.5 $4/$20 per 1M tokens, Sonnet 5.5 $2/$10, Haiku 5.5 $0.10/$0.50, Fable 5.1 $10/$50; Batch is 50% off. Pro is $20/month and Max starts at $100/month, both including Claude Code.
Claude Opus 5.5 input
Usage based4 USD / per 1M input tokens
Region
Global
Conditions
Cache hit $0.20; Batch 50% off; US-only inference 1.1x; Fast mode $8
Account access and model availability are determined by the provider. Subscription plans do not offset these API charges.
Standard synchronous text with uncached input; output includes billed thinking tokens. Excludes caching, batch, fast/priority, regional premiums, media, tools and taxes.
Account access and model availability are determined by the provider. Subscription plans do not offset these API charges.
Standard synchronous text with uncached input; output includes billed thinking tokens. Excludes caching, batch, fast/priority, regional premiums, media, tools and taxes.
Account access and model availability are determined by the provider. Subscription plans do not offset these API charges.
Standard synchronous text with uncached input; output includes billed thinking tokens. Excludes caching, batch, fast/priority, regional premiums, media, tools and taxes. Only the tier up to 100,000 input tokens. Larger inputs use different rates and are excluded.
Listed as an available legacy model. Check account access and retirement announcements. Continued availability is not guaranteed by this tool.
Anthropic first-party API, standard synchronous text-only requests. Excludes cache reads/writes, batch, third-party clouds, tools and taxes. Output must include all billed thinking tokens. Input plus output cannot exceed 200,000 tokens.