OpenAI model product
GPT-4
GPT-4 is a large multimodal model capable of processing both image and text inputs and generating human-like text outputs.
Updated Aug 12, 2026. Default version: GPT-4
LLMBoard score4.6GPT-4
Coverage80%8 benchmark families
Context window8.2KTokens
Official input price$30OpenAI API
GPT-4 Specifications
Technical details for the model's default version.
- Version
- GPT-4
- Released
- Jun 13, 2023
- Knowledge cutoff
- Dec 31, 2022
- Parameters
- N/A
- Context window
- 8.2K
- Max output
- 8.2K
- Inputs
- image, text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
GPT-4 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
GPT-4 Benchmark Results
Benchmark scores for GPT-4.
| AI2 Reasoning Challenge (ARC) | 96.3% | 01 | 1 | 100.0% | C | |
| LSAT | 88.0% | 01 | 1 | 100.0% | C | |
| SAT Math | 89.0% | 01 | 1 | 100.0% | C | |
| Uniform Bar Exam | 90.0% | 01 | 1 | 100.0% | C | |
| Winogrande | 87.5% | 01 | 22 | 100.0% | C | |
| HellaSwag | 95.3% | 02 | 27 | 96.2% | C | |
| DROP | 80.9% | 11 | 30 | 65.5% | C | |
| MGSM | 74.5% | 21 | 31 | 33.3% | C | |
| MMLU | 86.4% | 30 | 100 | 70.7% | C | |
| HumanEval | 67.0% | 59 | 66 | 10.8% | C | |
| MATH | 42.0% | 66 | 71 | 7.1% | C | |
| GPQA | 35.7% | 214 | 234 | 8.6% | C |
GPT-4 Arena Results
Preference and agent-evaluation results for the default version.
| text style control | overall | 244 | 1312.9 | 93,439 | N/A | |
| text style control | overall | 245 | 1312.5 | 100,105 | N/A | |
| text | overall | 256 | 1263.6 | 100,105 | N/A | |
| text | overall | 258 | 1262.3 | 93,439 | N/A | |
| text style control | overall | 266 | 1287.1 | 54,173 | N/A | |
| text style control | overall | 277 | 1275.4 | 88,723 | N/A | |
| text | overall | 287 | 1206.1 | 54,173 | N/A | |
| text | overall | 298 | 1186.1 | 88,723 | N/A |
GPT-4 Runtime Performance
Provider-specific output speed and catalog latency for GPT-4. Runtime does not affect the capability score.
| Azure | 104 tok/s | 0.3 s | 32.8K | 32.8K | |
| OpenAI | 100 tok/s | 0.5 s | 32.8K | 32.8K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
GPT-4 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $30 input, $60 output per 1M
- Official provider
- OpenAI
- Lowest third-party
- N/A
- Tracked offerings
- 1
| OpenAI | gpt-4 | global | $30 | $60 | 8.2K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
GPT-4 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| GPT-4 | 4.6 | N/A | 8.2K | 8.2K | No | Proprietary |
What is GPT-4?
Key information about GPT-4 and its available data.
GPT-4 is a large multimodal model capable of processing both image and text inputs and generating human-like text outputs. It demonstrates human-level performance on various professional and academic benchmarks.
Data as of 2026-08-11.
GPT-4 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to GPT-4
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
OP
GPT-4.1-nano
OpenAI
OP
GPT-4o-mini
OpenAI
OP
GPT-3.5-Turbo
OpenAI
OP
GPT-5.1-Codex-Mini
OpenAI
OP
GPT-4-Turbo
OpenAI
OP
o1 mini
OpenAI
FAQ
Common questions about GPT-4.
When was GPT-4 released?
GPT-4's default version was released on Jun 13, 2023.
How much does GPT-4 cost?
GPT-4's official API price is $30 per million input tokens and $60 per million output tokens via OpenAI.
Who created GPT-4?
GPT-4 was created by OpenAI.
What is the context window for GPT-4?
The default version has a 8.2K token context window.
Is GPT-4 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer GPT-4?
1 provider offerings are linked to the default version.
What models should I compare GPT-4 with?
Nearby ranked alternatives include Mistral Large 2, Qwen2.5 7B, Phi 3.5 MoE.