跳转到主要内容

GPT-5.2 (medium)

OpenAIGPT

描述

GPT‑5.2 introduces substantial gains in professional knowledge work, outperforming experts on GDPval with 70.9% wins or ties, and setting new highs in coding (SWE‑Bench Pro 55.6%), science (GPQA Diamond ~92–93%), math (AIME 2025: 100%), long‑context accuracy up to 256k tokens, and reliable tool‑calling (Tau2 Telecom 98.7%). It rolls out as Instant, Thinking, and Pro—faster, more structured, and less error‑prone—priced at $1.75/1M input and $14/1M output tokens, with Pro variants supporting xhigh reasoning for top‑quality, end‑to‑end execution.

发布日期
2025-12-11
参数规模
上下文长度
400K
支持模态
image, text

能力雷达图

50
general
80
coding
95
reasoning
60
science估算
70
agents
85
multimodal

Science 在缺少专门科学评测时使用推理能力代理估算。

排行榜排名

领域#排名分数来源
智能体能力模型榜44
58.0
LS
代码能力榜21
83.0
AA
通用能力榜56
74.0
AA
数学推理3
98.0
AA
多模态榜27
82.0
LS
推理能力30
82.0
LS
科学能力49
71.0
AA

基准测试分数 (LLM Stats)

Agents

BrowseComp65.8%自报
MCP Atlas60.6%自报
Toolathlon46.3%自报

Biology

GPQA92.4%自报

Code

SWE-Bench Verified80.0%自报
SWE-Lancer (IC-Diamond subset)74.6%自报

Communication

Tau2 Telecom98.7%自报
Tau2 Retail82.0%自报

General

MMMLU89.6%自报
MMMU-Pro79.5%自报
LiveBench74.8%自报

Grounding

ScreenSpot Pro86.3%自报

Healthcare

VideoMMMU85.9%自报

Math

AIME 2025100.0%自报
HMMT 202599.4%自报
FrontierMath40.3%自报
Humanity's Last Exam34.5%自报

Multimodal

CharXiv-R82.1%自报

Reasoning

Graphwalks BFS <128k94.0%自报
BrowseComp Long Context 128k92.0%自报
BrowseComp Long Context 256k89.8%自报
Graphwalks parents <128k89.0%自报
ARC-AGI86.2%自报
ARC-AGI v252.9%自报

AA 评测指数

Math Index
96.7
Intelligence Index
38.0
Aime 25
1.0
Livecodebench
0.9
Gpqa
0.9
Mmlu Pro
0.9
Tau2
0.7
Ifbench
0.7
Lcr
0.6
Scicode
0.5
Terminalbench Hard
0.4
Hle
0.2

LLM Stats 分类评分

Language
90
Physics
90
Grounding
90
Healthcare
90
Biology
90
Chemistry
90
Communication
90
Multimodal
80
Reasoning
80
Search
80
Spatial Reasoning
80
Frontend Development
80
General
80
Math
70
Code
70
Tool Calling
70
Vision
70
Agents
60

定价

输入价格$1.75 / 1M tokens
输出价格$14 / 1M tokens
混合价格(3:1)$4.813 / 1M tokens
缓存读取价格$0.175 / 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

供应商价格排行

17 个供应商

最便宜: QiHang最贵: Merge Gateway
供应商输入输出
1QiHang最便宜
$0.25
$2
2Jiekou.AI
$1.575
$12.6
3OpenAI主要
$1.75
$14
4302.AI
$1.75
$14
5Abacus
$1.75
$14
6OpenRouter
$1.75
$14
7Kilo Gateway
$1.75
$14
8GitHub Copilot
$1.75
$14
9OpenCode Zen
$1.75
$14
10AIHubMix
$1.75
$14
11Azure Cognitive Services
$1.75
$14
12Requesty
$1.75
$14
13Vercel AI Gateway
$1.75
$14
14LLM Gateway
$1.75
$14
15Azure
$1.75
$14
16OrcaRouter
$1.75
$14
17Merge Gateway
$1.75
$14

比较该模型在不同 API 供应商之间的定价。

外部链接