GPT-5.1 vs GPT-5.2 评测对比

See key specs and per-benchmark scores for each model/mode. Scroll horizontally for all columns. 当前对比 2 个模型的评测数据与核心参数。

GPT-5.1

OpenAI

Release: 2025-11-12
Context length: 400K
Parameters: Not provided
最大输出: 131,072 tokens
支持模态: 常规模式（Non-Thinking Mode） · 思考模式（Thinking Mode）

Model profile·Playground

Capability profile

Each axis is a category average, normalized to a 100-point radar.

View: Non-parallel mode average·6 dimensions

GPT-5.1

Relative edge: 常识推理 +7.4 / Relative gap: 综合评估 -12.0

GPT-5.2

Relative edge: 综合评估 +12.0 / Relative gap: 常识推理 -7.4

Method: for each model and benchmark, the chart first averages all scores in the current mode scope instead of taking the best score, then averages those benchmark scores within each category. Only benchmarks with at least two selected models scored are included; missing values are not counted as zero.

Best overall

GPT-5.2 · 67.81

Best single

GPT-5.2 · AIME2025 100.00

Modality coverage

GPT-5.1 · 2 modalities

Head to head

GPT-5.1

GPT-5.2

AheadTiedBehind

Benchmarks

Wins

Losses

-7.91

Average diff

Performance benchmarks

Compare benchmark results across thinking modes and tool usage.

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

Filter: Best Available·2 modes · 16 Benchmark

图表加载中...

Benchmark score table

Complete scores for each model/mode across selected benchmarks.

16 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.

Benchmark	GPT-5.1	GPT-5.2
ARC-AGI 综合评估	72.80Thinking Level · High	86.20Thinking Level · Extra High
ARC-AGI-2 综合评估	17.60Thinking Level · High	52.90Thinking Level · Extra High
GPQA Diamond 综合评估	88.10Thinking Enabled	92.40Thinking Level · Extra High
HLE 综合评估	42.70Thinking Level · High ｜ Tools	45.50Deep Thinking Mode ｜ Tools
LiveBench 综合评估	72.04Thinking Level · High	74.84Thinking Level · High
IC SWE-Lancer(Diamond) 编程与软件工程	69.70Thinking Level · High	74.60Thinking Level · Extra High ｜ Tools
SWE-Bench Pro - Public 编程与软件工程	50.80Thinking Level · High	55.60Thinking Level · Extra High ｜ Tools
SWE-bench Verified 编程与软件工程	76.30Thinking Level · High	80.00Thinking Level · Extra High ｜ Tools
AIME2025 数学推理	94.00Thinking Level · High	100.00Thinking Level · Extra High
FrontierMath 数学推理	26.70Thinking Level · High ｜ Tools	40.30Thinking Level · Extra High ｜ Tools
FrontierMath - Tier 4 数学推理	12.50Thinking Level · High ｜ Tools	18.80Thinking Level · Extra High
MMMU 多模态理解	85.40Thinking Level · High	85.90Thinking Level · Extra High
Simple Bench 常识推理	53.20Thinking Level · High	45.80Thinking Level · High
τ²-Bench - Telecom Agent能力评测	95.60Thinking Level · High ｜ Tools	98.70Thinking Level · Extra High ｜ Tools
BrowseComp AI Agent - 信息收集	50.80Thinking Level · High	65.80Thinking Level · Extra High ｜ Tools
MCP-Atlas AI Agent - 工具使用	50.10Thinking Level · High ｜ Tools	67.60Thinking Level · Extra High ｜ Tools

API price comparison

Side-by-side input/output token pricing

Detailed feature breakdown

Licensing, MoE architecture, and multi-modality support.

Features & specs	GPT-5.1OpenAI	GPT-5.2OpenAI
Core specsRelease	2025-11-12	2025-12-11
Context length	400K	400K
Max output	131072	Not provided
MoE	No	No
Supported modes	常规模式（Non-Thinking Mode）思考模式（Thinking Mode）	No mode data
LicenseCode Open Source	Not provided	Not provided
Weights Open Source	Not provided	Not provided
Commercial use	不开源	不开源
Modality supportText Input/Output	/	/
Image Input/Output	/	/
ResourcesPaper / report	GPT-5.1: A smarter, more conversational ChatGPT	Introducing GPT-5.2
DataLearner blog	OpenAI发布GPT-5.1：围绕“对话体验、一致性、任务适配性”进行的系统化优化的小幅更新！	Not provided