How we work

Methodology

Every number on this site comes from the provider's own pricing page. We don't use third-party aggregators or reseller prices.

What we record

• Standard-tier prices in US dollars per 1 million tokens: input, cached input (cache reads) and output.
• The context window, from the provider's model documentation.
• Conditions that change the price, such as surcharges for long prompts, peak-hour pricing or announced price changes. These appear in the Notes column.
• The date we last checked, shown on every page. The current data was checked on October 12, 2026.

How the calculator works

Monthly cost = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000.

Where a provider charges more for long prompts and publishes that rate, the calculator switches to it automatically once your input passes the threshold. Where the long-prompt rate isn't published, we use the standard rate and say so in the notes.

What the estimates leave out

• Prompt caching discounts, which can cut input costs a lot for repeated system prompts.
• Batch API discounts, usually 50% for jobs that can wait.
• Regional or data-residency surcharges, priority tiers and enterprise contracts.
• Reasoning or "thinking" tokens. Some models bill these as output, so real output can be higher than the visible answer.

Treat the numbers as a fair comparison between models, not as a quote.

Tokens in plain words

A token is a chunk of text. In English, 1,000 tokens is roughly 750 words. Different providers split text slightly differently, so the same prompt can count as a few percent more or fewer tokens from one model to the next.

Corrections

If you spot a wrong or outdated price, email [email protected] with a link to the provider's page. We fix confirmed errors quickly.

Independence

LLM Rate Card is not affiliated with any AI provider. If we ever add sponsored placements or affiliate links, they will be clearly labelled and will never change the order of results.