GPT-5 (high)

OpenAIGPTProprietary

描述

GPT-5 is a flagship model from OpenAI designed for coding, reasoning, and agentic tasks across domains. It is optimized for coding and agentic tasks with higher reasoning capabilities and medium speed.

发布日期

2025-08-07

参数规模

—

上下文长度

400K

支持模态

image, text

能力雷达图

general

coding

reasoning

science

agents

multimodal

排行榜排名

领域	#排名	分数	来源
智能体能力模型榜	75	45.0	LS
代码能力榜	134	66.0	AA
通用能力榜	68	76.0	AA
多模态榜	43	49.0	LS
科学能力	99	70.0	AA

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

Agents

BrowseCompOpenAI (2025)

54.9%自报

Biology

GPQANYU + Cohere + Anthropic (2023)

87.3%自报

Code

SWE-Lancer (IC-Diamond subset)

100.0%自报

HumanEvalOpenAI (2021)

93.4%自报

Aider-Polyglot

88.0%自报

SWE-Bench Verified

74.9%自报

Communication

Tau2 Telecom

96.7%自报

Tau2 Retail

81.1%自报

Multi-Challenge

69.6%自报

Tau2 Airline

62.6%自报

Finance

MMLU

92.5%自报

General

MMMU

84.2%自报

MMMU-Pro

78.4%自报

Internal API instruction following (hard)

64.0%自报

LongFact Objects

0.8%自报

LongFact Concepts

0.7%自报

Healthcare

VideoMMMU

84.6%自报

HealthBench Hard

1.6%自报

Language

COLLIE

99.0%自报

Long Context

OpenAI-MRCR: 2 needle 128k

95.2%自报

OpenAI-MRCR: 2 needle 256k

86.8%自报

Math

AIME 2025

94.6%自报

HMMT 2025

93.3%自报

MATH

84.7%自报

FrontierMath

26.3%自报

Humanity's Last Exam

24.8%自报

Multimodal

VideoMME w sub.

86.7%自报

CharXiv-R

81.1%自报

Reasoning

BrowseComp Long Context 128k

90.0%自报

BrowseComp Long Context 256k

88.8%自报

Graphwalks BFS <128k

78.3%自报

Graphwalks parents <128k

73.3%自报

ERQA

65.7%自报

FActScoreMin et al. (NYU/UW, 2023)

1.0%自报

AA 评测指数

(Artificial Analysis)

Math Index(Artificial Analysis)

94.3

Coding Index(Artificial Analysis)

37.8

Intelligence Index(Artificial Analysis)

35.3

Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))

1.0

Aime(MAA (Mathematical Association of America))

1.0

Aime 25(MAA (Mathematical Association of America))

0.9

Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))

0.9

Gpqa(NYU + Cohere + Anthropic (2023))

0.9

Tau2(Sierra + U Toronto + Vector Institute (2025))

0.8

Livecodebench(UC Berkeley + MIT + Cornell (2024))

0.8

Lcr(Artificial Analysis)

0.8

Ifbench(Google Research (2023))

0.7

Scicode(UIUC + Argonne National Lab (2024))

0.4

Terminalbench V2 1

0.4

Terminalbench Hard(Stanford × Laude Institute (2026))

0.3

Hle(Center for AI Safety + Scale AI (2025))

0.3

Tau Banking

0.2

LLM Stats 分类评分

(LLM Stats (zeroeval))

Spatial Reasoning

Vision

Reasoning

General

Long Context

100

Language

100

Writing

100

Legal

Physics

Finance

Biology

Chemistry

Code

Video

Multimodal

Communication

Tool Calling

Math

Frontend Development

Healthcare

Structured Output

Agents

定价

输入价格$1.25 / 1M tokens

输出价格$10 / 1M tokens

混合价格(3:1)$3.438 / 1M tokens

缓存读取价格$0.125 / 1M tokens

速度

Tokens/秒0.0

首Token延迟0.00s

首回答延迟0.00s

供应商价格排行

1 个供应商

供应商输入输出

1OpenAI主要

$1.25

$10

比较该模型在不同 API 供应商之间的定价。

外部链接

LLM Stats Artificial Analysis