Claude API pricing
Every Claude model we track, with the cheapest input and output rates per 1M tokens across 12 providers next to Anthropic's own API price. Collected daily.
Claude API prices by model
Standard (real-time) rates per 1M tokens, newest first. The cheapest input and the cheapest output are often on different hosts, so each is shown with its own host. Batch and cached-input rates are on each model page.
Official column shows input / output per 1M tokens on Anthropic's own API.
Claude price history
Average input price per 1M tokens across the providers serving each model, last 90 days.
Frequently Asked Questions
Which Claude model has the lowest API prices?
Of the 20 Claude models we track, Claude Haiku 5.5 has the lowest input rate right now: $0.100 per 1M tokens on IteraCompute. Claude Haiku 5.5 has the lowest output rate: $0.500 per 1M tokens on IteraCompute. Which is cheaper for you depends on how many input tokens you send per output token. Prices are collected daily from 12 providers.
Where is Claude Opus 5.5 cheapest?
Claude Opus 5.5 is served by 11 providers we track. The lowest rates are $2.00 per 1M input tokens on CloudGPU and $10.00 per 1M output tokens on CloudGPU. Throughput, latency and context limits differ between hosts, so check the model page before committing.
Is a third-party host cheaper than Anthropic's own API?
Of the 15 Claude models Anthropic serves directly, another provider currently lists a lower input rate for 11 and a lower output rate for 11. Compare the cheapest-input and cheapest-output columns with the "Official" column above; batch and cached-input rates are on each model page.