GPT-4o

Name: GPT-4o
Author: OpenAI

Multimodal

OpenAI

GPT-4o ('o' stands for 'omni') is a multimodal AI model that accepts text, audio, image, and video inputs and generates text, audio, and image outputs. It matches GPT-4 Turbo performance on text and code, with improvements in understanding non-English languages, images, and audio.

Key Specifications

Parameters

Context

128.0K

Release Date

August 6, 2024

Average Score

52.5%

API Documentation Results Blog

Timeline

Key dates in the model's history

Announcement

August 6, 2024

Last Update

July 19, 2025

Today

October 5, 2026

Technical Specifications

Parameters

Training Tokens

Knowledge Cutoff

Family

Capabilities

MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)

$2.50

Output (per 1M tokens)

$10.00

Max Input Tokens

128.0K

Max Output Tokens

16.4K

Supported Features

Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

General Knowledge

Tests on general knowledge and understanding

MMLU

Accuracy • Self-reported

85.7%

Programming

Programming skills tests

SWE-Bench Verified

Accuracy • Self-reported

33.2%

Reasoning

Logical reasoning and analysis

GPQA

GPT-4o - Diamond no thinking no tools • Self-reported

70.1%

Multimodal

Working with images and visual data

MathVista

Accuracy • Self-reported

61.4%

MMMU

GPT-4o without mode thinking - Solution visual tasks level with • Self-reported

72.2%

ChartQA

evaluation on test set • Self-reported

85.7%

DocVQA

evaluation on test set • Self-reported

92.8%

AI2D

Evaluation on test set AI: Translate full text • Self-reported

94.2%

Other Tests

Specialized benchmarks

MMLU-Pro

0-shot CoT • Self-reported

74.7%

SimpleQA

accuracy • Self-reported

38.2%

AIME 2024

Accuracy AI • Self-reported

13.1%

IFEval

Accuracy AI • Self-reported

81.0%

Aider-Polyglot

Accuracy AI • Self-reported

30.7%

EgoSchema

evaluation on test set • Self-reported

72.2%

Aider-Polyglot Edit

Accuracy AI • Self-reported

18.2%

MMMLU

Accuracy • Self-reported

81.4%

Multi-IF

Accuracy • Self-reported

60.9%

TAU-bench Retail

Accuracy • Self-reported

60.3%

TAU-bench Airline

Accuracy • Self-reported

42.8%

CharXiv-R

GPT-4o without mode thinking - justification and • Self-reported

58.8%

Internal API instruction following (hard)

Accuracy • Self-reported

29.2%

MultiChallenge (o3-mini grader)

Accuracy AI: translation text by evaluation accuracy models AI: Accuracy • Self-reported

39.9%

COLLIE

GPT-4o without mode thinking - instructions at text • Self-reported

61.0%

Tau2 airline

GPT-4o without mode thinking - Benchmark functions () • Self-reported

45.5%

OpenAI-MRCR: 2 needle 128k

Accuracy AI • Self-reported

31.9%

Tau2 retail

GPT-4o without mode thinking - Benchmark functions () • Self-reported

63.4%

Tau2 telecom

GPT-4o without mode thinking - Benchmark functions (field) • Self-reported

23.5%

MMMU-Pro

GPT-4o without mode thinking - Solution visual tasks level with reasoning • Self-reported

59.9%

VideoMMMU

GPT-4o without mode thinking - multimodal reasoning (256 ) • Self-reported

61.2%

ERQA

GPT-4o without mode thinking - thinking • Self-reported

35.2%

Graphwalks parents <128k

Accuracy • Self-reported

35.4%

CharXiv-D

Accuracy AI: PageRank connection how therefore with number receive more rating When receives with she/it receives part "" this is "" to • Self-reported

85.3%

ComplexFuncBench

Accuracy • Self-reported

66.5%

SWE-Lancer

result • Self-reported

32.6%

SWE-Lancer (IC-Diamond subset)

score • Self-reported

12.4%

Graphwalks BFS <128k

Accuracy • Self-reported

41.7%

ActivityNet

evaluation on test set • Self-reported

61.9%

Humanity's Last Exam

GPT-4o without mode thinking (without tools) - set questions expert level by various subjects • Self-reported

5.3%

Scale MultiChallenge

GPT-4o without mode thinking - Benchmark execution instructions • Self-reported

40.3%

Multi-Challenge

GPT-4o without thinking mode - Multi-turn instruction following benchmark. • Self-reported

40.3%

License & Metadata

License

proprietary

Announcement Date

August 6, 2024

Last Updated

July 19, 2025

Similar Models

All Models

GPT-4.1 mini

OpenAI

Best score:0.9 (MMLU)

Released:Apr 2025

Price:$0.40/1M tokens

o4-mini

OpenAI

Best score:0.8 (GPQA)

Released:Apr 2025

Price:$1.10/1M tokens

o3

OpenAI

Best score:0.8 (GPQA)

Released:Apr 2025

Price:$2.00/1M tokens

GPT-4o

OpenAI

Best score:0.9 (HumanEval)

Released:May 2024

Price:$2.50/1M tokens

GPT-4o mini

OpenAI

Best score:0.9 (HumanEval)

Released:Jul 2024

Price:$0.15/1M tokens

GPT-4.5

OpenAI

Best score:0.9 (MMLU)

Released:Feb 2025

Price:$75.00/1M tokens

GPT-4.1

OpenAI

Best score:0.9 (MMLU)

Released:Apr 2025

Price:$2.00/1M tokens

GPT-5 nano

OpenAI

Best score:0.7 (GPQA)

Released:Aug 2025

Price:$0.05/1M tokens

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.