Claude API pricing, checked at the source
Prices for Claude Fable, Opus, Sonnet and Haiku, read off Anthropic's own pricing page with a link on every row. Estimate a monthly bill below, with caching and batch.
Claude API prices per 1M tokens
Standard prices in USD. Each row links to Anthropic's pricing page and shows the day we checked it.
| Model | Input | Cached input | Output | Batch API (in / out) | Notes | Source |
|---|---|---|---|---|---|---|
| Claude Fable 5.1 | $10.00 | $0.25 | $50.00 | $5.00 / $25.00 | Cache 5m write $12.50, 1h write $20.00. | Sourcechecked 2026-10-25 |
| Claude Opus 5.5 | $4.00 | $0.20 | $20.00 | $2.00 / $10.00 | Fast mode: $8 input, $40 output.Cache 5m write $5.00, 1h write $8.00. | Sourcechecked 2026-10-25 |
| Claude Sonnet 5.5 | $2.00 | $0.10 | $10.00 | $1.00 / $5.00 | Cache 5m write $2.50, 1h write $4.00. | Sourcechecked 2026-10-25 |
| Claude Haiku 5.5prompts up to 100,000 tokens | $0.10 | $0.01 | $0.50 | $0.05 / $0.25 | Cache 5m write $0.125, 1h write $0.20. | Sourcechecked 2026-10-25 |
| Claude Haiku 5.5prompts over 100,000 tokens | $0.50 | $0.05 | $2.50 | $0.25 / $1.25 | Cache 5m write $0.625, 1h write $1.00. | Sourcechecked 2026-10-25 |
| Claude Fable 5 | $10.00 | $1.00 | $50.00 | $5.00 / $25.00 | Cache 5m write $12.50, 1h write $20.00. | Sourcechecked 2026-10-25 |
| Claude Opus 5 | $5.00 | $0.50 | $25.00 | $2.50 / $12.50 | Fast mode: $10 input, $50 output.Cache 5m write $6.25, 1h write $10.00. | Sourcechecked 2026-10-25 |
| Claude Sonnet 5 | $2.00 | $0.20 | $10.00 | $1.00 / $5.00 | Cache 5m write $2.50, 1h write $4.00. | Sourcechecked 2026-10-25 |
| Claude Sonnet 4.6 | $3.00 | $0.30 | $15.00 | $1.50 / $7.50 | Cache 5m write $3.75, 1h write $6.00. | Sourcechecked 2026-10-25 |
| Claude Haiku 4.5 | $1.00 | $0.10 | $5.00 | $0.50 / $2.50 | Cache 5m write $1.25, 1h write $2.00. | Sourcechecked 2026-10-25 |
Terms on this page: Token · Prompt caching · Context window · Batch API
Estimate your Anthropic bill
Enter your monthly tokens and how they split between new input, cached input and output. The result updates as you type.
Write 40k, 250M or 5B. Input, cached input and output tokens together.
100M tokens a month costs from $6.78 (Claude Haiku 5.5) to $678.00 (Claude Fable 5).
| Model | Per month | Per 1M tokens at this split |
|---|---|---|
| Claude Haiku 5.5prompts up to 100,000 tokens | $6.78 | $0.0678 |
| Claude Haiku 4.5 | $67.80 | $0.678 |
| Claude Sonnet 5.5 | $129.80 | $1.298 |
| Claude Sonnet 5 | $135.60 | $1.356 |
| Claude Sonnet 4.6 | $203.40 | $2.034 |
| Claude Opus 5.5 | $259.60 | $2.596 |
| Claude Opus 5 | $339.00 | $3.39 |
| Claude Fable 5.1 | $634.50 | $6.345 |
| Claude Fable 5 | $678.00 | $6.78 |
Standard prices in USD on 2026-10-25, before tax. Long-prompt tiers are left out; see the price table.
How the bill is worked out
Prompt caching
Batch API
Long context
Regions and tokenizer
Claude API pricing questions
How much does the Claude API cost?
Anthropic bills its language models per million tokens, with separate prices for input, cached input and output. On 2026-10-25, standard prices ran from $0.10 (Claude Haiku 5.5) to $10.00 (Claude Fable 5) per 1M input tokens, and $0.50 (Claude Haiku 5.5) to $50.00 (Claude Fable 5) per 1M output tokens.
Does Anthropic offer a batch discount?
Yes. Batch API costs 50% less than the standard price. It suits work that does not need an answer straight away, such as overnight document runs.
How does prompt caching change a Anthropic bill?
A cache hit costs 10% of the input price on most models, 5% on Opus 5.5 and Sonnet 5.5, and 2.5% on Fable 5.1. Writing to the cache costs 1.25x the input price for 5 minutes or 2x for 1 hour.
Where do these prices come from?
From Anthropic's own pricing page, https://platform.claude.com/docs/en/about-claude/pricing, checked on 2026-10-25. Every row links to it. When a price cannot be confirmed, the row says to check the vendor's page instead of showing a number.
Why does Claude Haiku 5.5 have two prices?
Haiku 5.5 is priced by prompt length. A request whose prompt is over 100,000 tokens pays the higher prices for the whole request, even when part of the prompt is a cache hit. Other Claude 4.6 and later models bill the full 1M-token context at the standard rate.
Do newer Claude models use more tokens for the same text?
Anthropic says Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text, so compare costs on your own prompts, not just the price per token.
Compare and plan
Building with the Claude API? OceanLab builds AI into business workflows, with people kept in charge. AI automation →
