GPT-4o (2024-05-13)
OpenAI · released May 13, 2024
GPT-4o ("o" for "omni") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...
Specification
- Context window
- 128K
- Max output
- 4K
- Knowledge cutoff
- Oct 31, 2023
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 2
- Moderated
- No
Intelligence
8.4
29th percentile
Coding
24.2
Coding Index
Agentic
—
Agentic Index
Output speed
78 t/s
Median across providers
Latency
602ms
Time to first token
Cost per task
$0.63
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- MMLU-Pro74.0%
- GPQA Diamond52.6%
- LiveCodeBench33.4%
- SciCode30.9%
- AIME 202511.0%
- Humanity's Last Exam1.8%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| MMLU-Pro | 74.0% |
| GPQA Diamond | 52.6% |
| LiveCodeBench | 33.4% |
| SciCode | 30.9% |
| AIME 2025 | 11.0% |
| Humanity's Last Exam | 1.8% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- SonarPerplexity9.4
- 9.4
- GPT-4o (2024-08-06)OpenAI9.4
- Sonar ProPerplexity9.1
- GPT-4o (2024-05-13)OpenAI8.4
- 8.2
- GPT-4 TurboOpenAI7.7
- Command ACohere7.5
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Sonar | 9.4 |
| Qwen2.5 72B Instruct | 9.4 |
| GPT-4o (2024-08-06) | 9.4 |
| Sonar Pro | 9.1 |
| GPT-4o (2024-05-13) | 8.4 |
| Qwen3 VL 8B Instruct | 8.2 |
| GPT-4 Turbo | 7.7 |
| Command A | 7.5 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $5
- Output / 1M tokens
- $15
- Cached input / 1M
- Not offered
- Blended 3:1
- $7.5
One task, estimated
$0.63
- Input tokens
- 50,000
- Output tokens
- 25,000
- Profile
- Standard
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Arena
Head-to-head generation quality
Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.
- Overall Elo
- 947
- Win rate
- 42.3%
- Strongest at
- Game development
- Tournaments
- 241
Elo by category
Dot position on a 850–950 scale · Elo has no meaningful zero
- 947
- 924
- 920
- 884
- 854
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| Game development | 947 | 42.3% |
| 3D scenes | 924 | 39.2% |
| UI components | 920 | 38.1% |
| Data visualisation | 884 | 36.0% |
| Websites | 854 | 31.5% |