OpenAI model product
GPT-4o
ChatGPT-4o Latest is a dynamic model that automatically points to the latest version of ChatGPT-4o.
Updated Aug 12, 2026. Default version: GPT-4o
LLMBoard score23.9GPT-4o
Coverage100%37 benchmark families
Context window128KTokens
Official input price$2.5OpenAI API
GPT-4o Specifications
Technical details for the model's default version.
- Version
- GPT-4o
- Released
- Aug 6, 2024
- Knowledge cutoff
- Sep 1, 2023
- Parameters
- N/A
- Context window
- 128K
- Max output
- 16.4K
- Inputs
- image, text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
GPT-4o Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
GPT-4o Benchmark Results
Benchmark scores for GPT-4o.
| ActivityNet | 61.9% | 01 | 1 | 100.0% | C | |
| ComplexFuncBench | 66.5% | 01 | 7 | 100.0% | C | |
| AI2D | 94.2% | 03 | 32 | 93.5% | C | |
| EgoSchema | 72.2% | 03 | 9 | 75.0% | C | |
| SWE-Lancer | 32.6% | 03 | 4 | 33.3% | C | |
| SWE-Lancer (IC-Diamond subset) | 12.4% | 05 | 6 | 20.0% | C | |
| Internal API instruction following (hard) | 29.2% | 07 | 7 | 0.0% | C | |
| COLLIE | 61.0% | 08 | 10 | 22.2% | C | |
| OpenAI-MRCR: 2 needle 128k | 31.9% | 08 | 9 | 12.5% | C | |
| Aider-Polyglot Edit | 18.2% | 09 | 10 | 11.1% | C | |
| Graphwalks BFS <128k | 41.7% | 10 | 11 | 10.0% | C | |
| Graphwalks parents <128k | 35.4% | 10 | 11 | 10.0% | C | |
| ChartQA | 85.7% | 12 | 24 | 52.2% | C | |
| CharXiv-D | 85.3% | 12 | 16 | 26.7% | C | |
| DocVQA | 92.8% | 15 | 26 | 44.0% | C | |
| Multi-IF | 60.9% | 19 | 20 | 5.3% | C | |
| TAU-bench Airline | 42.8% | 19 | 23 | 18.2% | C | |
| Aider-Polyglot | 30.7% | 20 | 22 | 9.5% | C | |
| TAU-bench Retail | 60.3% | 20 | 25 | 20.8% | C | |
| SimpleQA | 38.2% | 21 | 46 | 55.6% | C | |
| Tau2 Airline | 45.5% | 21 | 23 | 9.1% | C | |
| Multi-Challenge | 40.3% | 22 | 29 | 25.0% | C | |
| ERQA | 35.2% | 23 | 23 | 0.0% | C | |
| Tau2 Retail | 63.4% | 23 | 26 | 12.0% | C | |
| VideoMMMU | 61.2% | 25 | 26 | 4.0% | C | |
| MathVista | 61.4% | 26 | 39 | 34.2% | C | |
| MMMU | 72.2% | 28 | 63 | 56.5% | C | |
| Tau2 Telecom | 23.5% | 34 | 35 | 2.9% | C | |
| CharXiv-R | 58.8% | 37 | 48 | 23.4% | C | |
| MMLU | 85.7% | 37 | 100 | 63.6% | C | |
| MMMLU | 81.4% | 38 | 49 | 22.9% | C | |
| IFEval | 81.0% | 51 | 65 | 21.9% | C | |
| MMMU-Pro | 59.9% | 51 | 66 | 23.1% | C | |
| AIME 2024 | 13.1% | 53 | 53 | 0.0% | C | |
| MMLU-Pro | 74.7% | 72 | 129 | 44.5% | C | |
| Humanity's Last Exam | 5.3% | 88 | 93 | 5.4% | C | |
| SWE-Bench Verified | 33.2% | 100 | 105 | 4.8% | C | |
| GPQA | 70.1% | 128 | 234 | 45.5% | C |
GPT-4o Arena Results
Preference and agent-evaluation results for the default version.
| vision style control | overall | 112 | 1119.4 | 3,376 | N/A | |
| vision | overall | 116 | 1064.7 | 3,376 | N/A | |
| text style control | overall | 214 | 1335.3 | 45,499 | N/A | |
| text | overall | 237 | 1282.7 | 45,499 | N/A |
GPT-4o Runtime Performance
Provider-specific output speed and catalog latency for GPT-4o. Runtime does not affect the capability score.
| OpenAI | 184.314 tok/s | 1.096 s | 128K | 16.4K | |
| Azure | 99 tok/s | 0.53 s | 128K | 16.4K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
GPT-4o Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $2.5 input, $10 output per 1M
- Official provider
- OpenAI
- Lowest third-party
- From $2.5 input, $10 output per 1M via NanoGPT
- Tracked offerings
- 6
| NanoGPT | openai/gpt-4o-2024-08-06 | global | $2.5 | $10 | 128K | |
| OrcaRouter | openai/gpt-4o-2024-08-06 | global | $2.5 | $10 | 128K | |
| Kilo Gateway | openai/gpt-4o-2024-08-06 | global | $2.5 | $10 | 128K | |
| Merge Gateway | openai/gpt-4o-2024-08-06 | global | $2.5 | $10 | 128K | |
| OpenRouter | openai/gpt-4o-2024-08-06 | global | $2.5 | $10 | 128K | |
| OpenAI | gpt-4o-2024-08-06 | global | $2.5 | $10 | 128K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
GPT-4o Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| GPT-4o | 23.9 | N/A | 128K | 16.4K | No | Proprietary | |
| ChatGPT-4o Latest | N/A | N/A | 128K | 16.4K | No | Proprietary | |
| GPT-4o | N/A | N/A | 128K | 4.1K | No | Proprietary |
What is GPT-4o?
Key information about GPT-4o and its available data.
ChatGPT-4o Latest is a dynamic model that automatically points to the latest version of ChatGPT-4o. This ensures you always have access to the most recent improvements and capabilities without changing your integration.
Data as of 2026-08-11.
GPT-4o vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to GPT-4o
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
OP
GPT-4.1-mini
OpenAI
OP
o1 mini
OpenAI
OP
GPT-5-nano
OpenAI
OP
GPT-4.1
OpenAI
OP
GPT-4-Turbo
OpenAI
OP
o1 pro
OpenAI
FAQ
Common questions about GPT-4o.
When was GPT-4o released?
GPT-4o's default version was released on Aug 6, 2024.
How much does GPT-4o cost?
GPT-4o's official API price is $2.5 per million input tokens and $10 per million output tokens via OpenAI. The lowest tracked third-party offer starts at $2.5 input and $10 output via NanoGPT.
Who created GPT-4o?
GPT-4o was created by OpenAI.
What is the context window for GPT-4o?
The default version has a 128K token context window.
Is GPT-4o open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer GPT-4o?
6 provider offerings are linked to the default version.
What models should I compare GPT-4o with?
Nearby ranked alternatives include Llama 3.1 405B, Gemma 4 E4B, Devstral Medium.