GPT-5

Name: GPT-5
Author: OpenAI

Foundation modelGPTGPT-5

GPT-5

Release date: 2025-08-07Updated: 2026-06-15 07:18:15.2672,572

Live demoGitHubHugging FaceCompare

Parameters

Not disclosed

Context length

400K

Chinese support

Supported

Reasoning ability

GPT-5 is an AI model published by OpenAI, released on 2025-08-07, for Foundation model, and 400K context length, under the 不开源 license, with a 1360.00 score on CodeClash.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

GPT-5

Model basics

Reasoning traces

Supported

Thinking modes

Thinking modes not supported

Context length

400K tokens

Max output length

128K tokens

Model type

Foundation model

Modality (in / out)

Text, Image → Text

Release date

2025-08-07

Model file size

No data

MoE architecture

Total params / Active params

No data / N/A

Knowledge cutoff

No data

GPT-5

Open source & experience

Code license

不开源

Weights license

不开源

GitHub repo

GitHub link unavailable

Hugging Face

Hugging Face link unavailable

Live demo

https://chat.openai.com/

GPT-5

Official resources

Paper

Introducing GPT-5

DataLearnerAI blog

GPT-5

API details

API speed

3/5

No public API pricing yet.

GPT-5

Benchmark Results

GPT-5 currently shows benchmark results led by Aider-Polyglot (1 / 59, score 88), AIME2025 (9 / 106, score 99.60), IMO-ProofBench (2 / 16, score 59). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

General Knowledge

14 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamond

87.30

38 / 179

GPQA Diamond

85.70

45 / 179

GPQA Diamond

77.80

85 / 179

ARC-AGI

65.70

30 / 65

ARC-AGI

56.20

40 / 65

ARC-AGI

45 / 65

ARC-AGI

61 / 65

HLE

35.20

62 / 159

HLE

24.80

90 / 159

HLE

6.30

148 / 159

ARC-AGI-2

9.90

37 / 59

ARC-AGI-2

7.50

40 / 59

ARC-AGI-2

1.90

50 / 59

ARC-AGI-2

57 / 59

Coding and Software Engineer

3 evaluations

Benchmark / mode

Score

Rank/total

CodeClash

Standard ModeTools

1360

2 / 8

SWE-bench Verified

72.80

46 / 108

SWE-Bench Pro - Public

36.30

42 / 44

Math and Reasoning

12 evaluations

Benchmark / mode

Score

Rank/total

AIME2025

99.60

9 / 106

AIME2025

94.60

26 / 106

AIME2025

61.90

80 / 106

IMO-ProofBench

2 / 16

IMO 2025

2 / 9

FrontierMath

24.80

15 / 60

FrontierMath

24.80

15 / 60

FrontierMath

HighTools

26.30

14 / 60

IMO-ProofBench Advanced

2 / 8

FrontierMath - Tier 4

Medium

6.30

35 / 80

FrontierMath - Tier 4

High

12.50

29 / 80

IMO 2024

4 / 10

AI Agent - Tool Usage

1 evaluations

Benchmark / mode

Score

Rank/total

Terminal-Bench

43.80

8 / 35

Multimodal Understanding

1 evaluations

Benchmark / mode

Score

Rank/total

MMMU

84.20

5 / 28

常识推理

1 evaluations

Benchmark / mode

Score

Rank/total

Simple Bench

High

56.70

20 / 63

Agent Level Benchmark

6 evaluations

Benchmark / mode

Score

Rank/total

τ²-Bench - Telecom

95.80

13 / 35

τ²-Bench - Telecom

HighTools

96.70

11 / 35

Aider-Polyglot

Low

81.30

5 / 59

Aider-Polyglot

Medium

86.70

2 / 59

Aider-Polyglot

High

1 / 59

τ²-Bench

15 / 40

Instruction Following

1 evaluations

Benchmark / mode

Score

Rank/total

IF Bench

73.10

8 / 29

AI Agent - Information Search

1 evaluations

Benchmark / mode

Score

Rank/total

BrowseComp

54.90

32 / 45

View benchmark analysis Compare with other models

Compare with other models

Want a custom combination? Open the compare tool

GPT-5

Publisher

OpenAI

View publisher details

GPT-5

Model Overview

GPT-5 is an AI model published by OpenAI, released on 2025-08-07, for Foundation model, and 400K context length, under the 不开源 license, with a 1360.00 score on CodeClash.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.