Claude API Pricing (2026): Fable 5.2, Opus 5.5, Sonnet 5.5 and Haiku 5.5 Token Prices, Prompt Caching, Batch and Fast Mode

Last updated: 13 November 2026. All prices checked on the vendors’ official pages that day.

Claude API pricing is per million tokens (MTok), with separate rates for input, output, cache writes and cache reads. Anthropic’s current lineup in November 2026 is Claude Fable 5.2 for demanding reasoning and long-horizon agents, Claude Opus 5.5 (Anthropic’s suggested starting point for most workloads), Claude Sonnet 5.5 and the very low-cost Claude Haiku 5.5. This page lists the official prices, the modifiers that move them (caching, Batch, Fast mode, data residency) and what typical workloads cost.

Source: Anthropic’s Claude API pricing documentation, checked 13 November 2026. To compare Claude with GPT and Gemini, see LLM API pricing comparison.

Claude API prices per 1M tokens

• Claude Mythos 5.2 has the same list price as Fable 5.2 ($8 / $40) but is limited to organisations verified through Anthropic programmes such as the Cyber Verification Program.
• Sonnet 5 stays at $2 / $10: Anthropic made the introductory price permanent and cancelled the planned rise to $3 / $15.
• Older models still listed: Opus 5, 4.8, 4.7, 4.6 and 4.5 at $5 / $25; Sonnet 4.6 and 4.5 at $3 / $15; Haiku 4.5 at $1 / $5.
• Haiku 5.5 has two prices. A request whose prompt (including cache reads and writes) is over 100,000 tokens pays the higher rate for that whole request.

Long context: 1M tokens at the standard rate

Claude 4.6 and later models (except Haiku 5.5) include the full 1M-token context window at standard pricing. Anthropic’s own example: a 900K-token request is billed at the same per-token rate as a 9K-token request. That is a real difference from vendors that add a long-context surcharge.

Prompt caching, Batch and Fast mode

• Cache writes cost 1.25× base input for the 5-minute cache and 2× for the 1-hour cache.
• Cache hits cost 0.1× base input on most models, 0.05× on Opus 5.5 and Sonnet 5.5, and 0.025× on Fable 5.2/5.1 and Mythos 5.2/5.1.
• Batch API: 50% off input and output. It stacks with caching.
• Fast mode (up to 2.5× faster output) is a premium: Fable 5.2 and Mythos 5.2 at $16 / $80, Opus 5.5 at $8 / $40, Opus 5 and 4.8 at $10 / $50. Not available with Batch.
• Data residency: setting inference_geo: "us" on Claude 4.6 and later applies a 1.1× multiplier to all token categories.

Tokenizer change: count again

Claude Opus 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text. If you migrate from Sonnet 4.6 to a 5.x model, recount your prompts with the token counting endpoint instead of reusing old numbers. A lower per-token price can be partly offset by more tokens. More on this in what is a token in AI.

Other Claude API charges

• Web search: $10 per 1,000 searches plus the search results as input tokens.
• Code execution is free when used together with the latest web search or web fetch tools.
• Claude Managed Agents: tokens at model rates plus session runtime charges.

Claude API cost examples

How to pay and where it is available

Claude API billing is based on actual monthly usage in USD. Standard accounts buy credits by card on the Claude Console billing page; enterprise customers can arrange invoicing. New users receive some free credits for testing. The API is only offered in Anthropic’s supported countries and regions, which include Singapore and Taiwan but not mainland China or Hong Kong. Step-by-step setup: how to get a Claude API key.

PandaWM helps companies open and manage their own Claude Console organization through an entity that is legitimately established in a supported region, set up workspaces, keys and spend limits, or connect Claude alongside GPT and Gemini through a compliant unified API: Overseas AI Token service: account setup + API access (PandaWM). See also OpenAI API pricing and Gemini API pricing.

FAQ

How much does Claude Sonnet 5.5 cost?

$2 per 1M input tokens and $10 per 1M output tokens; cache hits are $0.10 and Batch halves both rates.

What is the cheapest Claude model?

Claude Haiku 5.5 at $0.10 input and $0.50 output per 1M tokens for prompts up to 100,000 tokens.

Does Claude charge extra for long prompts?

Not on Claude 4.6 and later models other than Haiku 5.5: the 1M context window is billed at standard rates.

Is a Claude Pro subscription the same as the API?

No. Claude app plans and the Claude API are billed separately; API usage is billed in the Claude Console.

Can we combine Batch and caching?

Yes. Anthropic states the Batch discount and prompt caching multipliers stack.

Is PandaWM an Anthropic reseller?

No. We help companies use their own accounts or a compliant unified API. Prices shown are Anthropic’s list prices.

Compliance note: Each model vendor only serves customers in its supported countries and regions and requires accurate account, identity and billing information. PandaWM helps companies use AI models through their own accounts or overseas entities that are legitimately established in supported regions, or through a compliant unified API. We do not sell, rent, share or transfer accounts or API keys, and we do not help anyone bypass regional restrictions, KYC, sanctions or payment controls. We are not an official partner or reseller of OpenAI, Anthropic, Google or any other vendor. Prices on this page are the vendors’ public list prices, not PandaWM prices.

Related guides

• Overseas AI Token service: account setup + API access (PandaWM)
• OpenAI API pricing (GPT-6) + cost calculator
• ChatGPT API pricing vs ChatGPT plans
• Gemini API pricing, free tier and billing tiers
• LLM API pricing comparison: cheapest providers
• How to get a Claude (Anthropic) API key
• What is a token in AI? Token cost explained
• 简体中文：大模型 API 价格对比与 Token 计算器

Official sources

• Anthropic: Claude API pricing
• Anthropic: token counting
• Anthropic: supported regions
• Anthropic: get your Claude API key

All official pages checked on 2026-11-13. Prices are in USD per 1 million tokens (MTok), before tax. Vendors change prices; the official pages prevail.