Compare token pricing across Claude Fable 5, Opus 4.8, Sonnet 5, Haiku 4.5, GPT-5.6 and Gemini 3.1 Pro — per call, per day, per month.
✓ Prices verified July 9, 2026| Model | $/M in · out | Per call | Per day | Per month |
|---|
How accurate is this?
Prices are the official published API rates as of July 9, 2026. Real bills vary with prompt caching (up to 90% off cached input), batch discounts (50%), and always-on thinking tokens on Fable 5.
Why is Sonnet 5 shown twice?
Anthropic's introductory pricing ($2/$10) runs through August 31, 2026; the standard rate is $3/$15. Both are shown so you can budget past September.
Why is GPT-5.6 marked "restricted"?
As of July 2026 it is limited to ~20 government-approved partners. Prices are vendor-published, included for comparison only.
Tokens vs words?
Budget roughly 750 English words (or ~2,500 characters of code) per 1,000 tokens.