GPT-4o API Pricing
Deprecated: on October 14, 2026 OpenAI told developers that GPT-4o - the gpt-4o-2024-08-06 and gpt-4o-2024-11-20 snapshots behind the gpt-4o alias - leaves the API on January 14, 2027, and it names GPT-6 Luna as the replacement. Until then it bills $2.50/M input, $1.25/M cached input and $10.00/M output, an effective $2.16/M on this site's agentic blend. The older gpt-4o-2024-05-13 snapshot, which kept the $5/$15 launch price, was shut down on October 23, 2026.
GPT-4o API cost calculator.
Rates are the current ones for the gpt-4o alias, not the frozen 2024-05-13 snapshot. Cache misses bill at the full input rate here: OpenAI publishes no cache-write charge for this model, so there is no separate write line to model.
GPT-4o cost on real workloads.
Repo-wide bug fix
Reading 100-page contracts
Support agent ticket triage
Research planning turn
This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| GPT-4o Current | $2.50 cache $1.25 | $10.00 | $2.16 agentic 92/8 | 128K | Legacy multimodal apps already wired to it |
| GPT-5.4 mini | $0.75 cache $0.075 | $4.50 | $0.541 −4.0× | 400K | The honest replacement - newer and a quarter the blend |
| GPT-5.4 | $2.50 cache $0.25 | $15.00 | $1.80 −16% | 1.05M | Same input rate, newer model, bigger context |
| GPT-5.5 | $5.00 cache $0.50 | $30.00 | $3.61 +67% | 1M | Named as the migration target, but costs more here |
| Claude Sonnet 4.6 | $3.00 cache $0.30 | $15.00 | $2.05 −5% | 1M | Legacy Claude production baselines |
| Gemini 3.8 Flash | $0.75 cache $0.075 | $3.75 | $0.481 −4.5× · promo | 1M | Gemini long-context work |
| DeepSeek V4 Pro | $1.32 cache $0.044 | $3.96 | $0.569 −3.8× | 1M | Low-cost reasoning and coding |
GPT-4o has published scores on 2 benchmarks we track; its best placing is #97 of 105 on ARC-AGI-2. All of these scores come from independent leaderboards, not from the lab.
| Benchmark | Score | Rank | Leader | Source |
|---|---|---|---|---|
| ARC-AGI-2 Reasoning | 0% ARC Prize verified, best result | #97 of 105 | GPT-6 Astra · 95% | Official leaderboard ARC Prize Foundation |
| ARC-AGI-1 Reasoning | 4.5% ARC Prize verified, best result | #95 of 99 | 6 tied · 98.5% | Official leaderboard ARC Prize Foundation |
GPT-4o API pricing FAQ.
What GPT-4o costs until it leaves the API, why older pages quote $5/$15, and what OpenAI says to move to.
Q · 01 Is GPT-4o still available? +
gpt-4o-2024-08-06 and gpt-4o-2024-11-20 - is removed from the API on that date, and its model page now marks every snapshot behind gpt-4o Deprecated or Shut down. Until then the alias bills at the rates on this page. Before that notice this page said the alias was on no retirement table, which was true from our September 6 check until October 14; we found the notice on October 28.Q · 02 How much does GPT-4o cost? +
Q · 03 Why do I see $5 and $15 quoted elsewhere? +
gpt-4o-2024-08-06 release — its Structured Outputs announcement says developers save 50% on inputs ($2.50/1M input tokens) and 33% on outputs ($10.00/1M output tokens) compared to gpt-4o-2024-05-13. The dated snapshot kept the old rates until it was shut down on October 23, 2026. Any page quoting $5/$15 for plain gpt-4o is two years out of date; this one was, until September 6, 2026.Q · 04 Is cached input a 90% discount like the newer models? +
Q · 05 What should teams use instead? +
gpt-6-luna as the replacement for GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini alike. Luna lists at $0.10/M input and $0.50/M output, with a 1.05M context against 128K here, and blends to $0.0682/M against $2.16/M. If you need more capability than Luna, the shelf below prices the larger GPT-6 tiers on the same blend.