gpt-4o
GPT-4o — priced by OpenAI.
Published tiers
What the most recent complete run published for this model, one row per priced tier, each with the source it was read from.
| Tier | Input USD/1M | Output USD/1M | Last read | Read from | Flags |
|---|---|---|---|---|---|
| batch | 1.25 | 5 | OpenAI API pricing (docs) | confirmedcorroborated | |
| cache_hit | 1.25 | n/a, not applicable: this tier prices input tokens only and has no output side | OpenAI API pricing (docs) | confirmedcorroborated | |
| cache_hit:batch | 0.625 | n/a, not applicable: this tier prices input tokens only and has no output side | OpenRouter models API | confirmedsingle source | |
| cache_hit:priority | 2.125 | n/a, not applicable: this tier prices input tokens only and has no output side | OpenAI API pricing (docs) | confirmedsingle source | |
| flat | 2.5 | 10 | OpenAI API pricing (docs) | confirmedcorroborated | |
| flat:priority | 4.25 | 17 | OpenAI API pricing (docs) | confirmedsingle source |
The same six decades as the front page, so a price is in the same place here as it is anywhere else on this site. The two marks in the accent colour are this model’s cheapest and dearest published values.
Price history
Every distinct value this price has taken, oldest first — what the price did, not how often we happened to look. Rows marked backfill came from a public git history of a third-party pricing file and were never verified by us; they extend the series backwards and are flagged rather than hidden.
batch
1 distinct valueOne value across the whole archive: this price has not moved in anything we have observed.
Held steady662 days — the value published now has been observed to hold for 662 days, which is 89% of the longest run on this page.
cache_hit
1 distinct valueOne value across the whole archive: this price has not moved in anything we have observed.
Held steady743 days — the value published now has been observed to hold for 743 days, which is 100% of the longest run on this page.
cache_hit:batch
1 distinct valueOne value across the whole archive: this price has not moved in anything we have observed.
Held steady63 days — the value published now has been observed to hold for 63 days, which is 8% of the longest run on this page.
cache_hit:priority
1 distinct valueOne value across the whole archive: this price has not moved in anything we have observed.
Held steady62 days — the value published now has been observed to hold for 62 days, which is 8% of the longest run on this page.
flat
2 distinct valuesHeld steady703 days — the value published now has been observed to hold for 703 days, which is 95% of the longest run on this page.
inputoutput1 window nobody looked inhorizontal axis is time
| First seen | Last seen | Input USD/1M | Output USD/1M | Reads | Source | Last read | Flags |
|---|---|---|---|---|---|---|---|
| 2024-05-13 18:12Z | 2024-05-13 18:12Z | 5 | 15 | 1 | LiteLLM model_prices_and_context_window.json (git history) | never | backfillunconfirmed |
| 2024-11-13 02:40Z | 2026-10-17 12:00Z | 2.5 | 10 | 74 | OpenAI API pricing (docs) | 2026-10-17 12:00Z |
flat:priority
1 distinct valueOne value across the whole archive: this price has not moved in anything we have observed.
Held steady62 days — the value published now has been observed to hold for 62 days, which is 8% of the longest run on this page.
Provenance
Read by openai_astro_props (batch, cache_hit, cache_hit:priority, flat, flat:priority) and openrouter_json (cache_hit:batch) from OpenAI API pricing (docs), OpenRouter models API. Each row above links to the exact page its number came from. The vendor's page is authoritative: if it and this page disagree, the vendor is right and we are wrong.