How much does the Claude API cost?

Current model prices

• Fable 5.1: $10 input, $50 output, $12.50 cache write and $0.25 cache read per million tokens, per Anthropic's pricing page checked Oct. 28, 2026.
• Opus 5.5: $4 input, $20 output, $5 cache write and $0.20 cache read per million tokens, per the same page on Oct. 28, 2026.
• Sonnet 5.5: $2 input, $10 output, $2.50 cache write and $0.10 cache read per million tokens, per the same page on Oct. 28, 2026.
• Haiku 5.5 on prompts up to 100K tokens: $0.10 input, $0.50 output, $0.125 cache write and $0.01 cache read per million tokens, per the same page on Oct. 28, 2026.
• Haiku 5.5 on prompts over 100K tokens: $0.50 input, $2.50 output, $0.625 cache write and $0.05 cache read per million tokens, per the same page on Oct. 28, 2026.

Legacy model prices

• Fable 5: $10 input, $50 output, $12.50 cache write and $1 cache read per million tokens, per Anthropic's pricing page checked Oct. 28, 2026.
• Opus 5, Opus 4.8, Opus 4.7, Opus 4.6 and Opus 4.5: $5 input, $25 output, $6.25 cache write and $0.50 cache read per million tokens, per the same page on Oct. 28, 2026.
• Sonnet 5: $2 input, $10 output, $2.50 cache write and $0.20 cache read per million tokens, per the same page on Oct. 28, 2026.
• Sonnet 4.6 and Sonnet 4.5: $3 input, $15 output, $3.75 cache write and $0.30 cache read per million tokens, per the same page on Oct. 28, 2026.
• Haiku 4.5: $1 input, $5 output, $1.25 cache write and $0.10 cache read per million tokens, per the same page on Oct. 28, 2026.

What else changes the bill

• Batch processing saves 50%, per Anthropic's pricing page checked Oct. 28, 2026. Anthropic's batch docs, checked the same day, say most batches finish in less than 1 hour.
• Cache prices on the pricing page reflect a 5 minute cache lifetime. Anthropic's pricing docs, checked Oct. 28, 2026, list 1 hour cache writes at $8 per million tokens for Opus 5.5 and $4 for Sonnet 5.5.
• US-only inference costs 1.1 times the standard input and output price, per Anthropic's pricing page checked Oct. 28, 2026.
• Fast mode for Opus 5.5 costs 2 times the standard price for up to 2.5 times faster output, per the same page on Oct. 28, 2026.
• Web search costs $10 per 1,000 searches plus the tokens used, per the same page on Oct. 28, 2026.
• Code execution includes 50 free hours a day per organization, then $0.05 per hour per container, per the same page on Oct. 28, 2026.
• Managed Agents cost $0.08 per session hour of active runtime plus standard token rates, per the same page on Oct. 28, 2026.

How to estimate your monthly cost

1. Measure the average input and output tokens of one real request in your product.
2. Multiply each by your expected monthly request count and divide by one million.
3. Multiply input and output millions by the model's rates. At the Oct. 28, 2026 Sonnet 5.5 rates, 1,000 requests of 2,000 input and 500 output tokens cost $4 for input plus $5 for output, or $9.
4. Move repeated system prompts into the cache and send work that can wait through the batch API, then recalculate.

Sources

1. Pricing · Anthropic, Oct. 28, 2026Supports: Checked Oct. 28, 2026 for every model price, the 50% batch saving, US-only inference, fast mode, web search, code execution and Managed Agents fees.
2. Pricing (Claude Platform docs) · Anthropic, Oct. 28, 2026Supports: Checked Oct. 28, 2026 for 1 hour cache write prices and the batch discount on input and output tokens.
3. Batch processing · Anthropic, Oct. 28, 2026Supports: Checked Oct. 28, 2026 for the 50% batch saving and the note that most batches finish in less than 1 hour.

Related questions

• What is an AI wrapper?
• Is Claude HIPAA compliant?

Intrica helps AI companies get found in Google and named in AI answers. SEO and AI search for AI companies →