AI Signal

Verified data

What the major LLM APIs actually cost

Input and output pricing for 72 current models across 9 providers. Every figure was read off the provider's own pricing page, and every row carries the date it was checked and a link to the page it came from — so you can verify any number here in one click instead of taking our word for it.

Model prices verified between September 11 and October 14, 2026; 3 of 72 models still carry the oldest date (verified includes confirming a provider publishes no price) · How the data is checked

Pricing table →

All 72 current models side by side: input, output, cached input, context window, and the source for each. Filter by provider. 4 say "no published price", because the maker publishes no per-token price.

Cost calculator →

Put in your monthly token volume and see what each model would actually bill you, instead of comparing rates per million in your head.

Price changes →

Every change to a price on this site, dated and sourced: 6 by providers, 18 of this site's own corrections, kept apart.

How pricing works →

Tokens, cache, long-prompt tiers, Batch, region, peak hours, promotions and hosts: what sets the price of a call, each shown on a real row.

New models →

LLM APIs by release date, with their price and the day each entered this table. Newest: Gemini 4 Argon.

Deprecations →

Model shutdown dates and the replacement each provider names, quoted from its deprecations page. Next: Google, October 22, 2026.

Methodology →

How each price is read on the maker's own page, re-checked weekly against a verbatim quote, reviewed by hand, and every correction recorded with its date.

Analysis

What the dataset shows when it is read across providers.