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
OpenAI
Each axis is a category average, normalized to a 100-point radar.
Relative edge: AI Agent - 工具使用 +18.5 / Relative gap: 常识推理 -5.7
Relative edge: 常识推理 +5.7 / Relative gap: AI Agent - 工具使用 -18.5
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 · 53.53
Best single
GPT-5 · AIME2025 99.60
Modality coverage
GPT-5 · 2 modalities
Head to head
19
Benchmarks
17
Wins
2
Losses
+12.60
Average diff
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
Complete scores for each model/mode across selected benchmarks.
19 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.
| Benchmark | GPT-5 | Gemini 2.5-Pro |
|---|---|---|
ARC-AGI 综合评估 | 65.70Thinking Level · High | 37.00Thinking Enabled |
ARC-AGI-2 综合评估 | 9.90Thinking Level · High | 4.90Thinking Enabled |
GPQA Diamond 综合评估 | 87.30Thinking Enabled ｜ Tools | 86.40Thinking Enabled |
HLE 综合评估 | 35.20Thinking Enabled ｜ Tools | 21.60Thinking Enabled |
LiveBench 综合评估 | 79.33Thinking Level · High | 71.92Thinking Enabled |
SWE-bench Verified 编程与软件工程 | 72.80Thinking Level · High | 67.20Thinking Enabled |
AIME2025 数学推理 | 99.60Thinking Enabled ｜ Tools | 88.00Thinking Enabled |
FrontierMath 数学推理 | 26.30Thinking Level · High ｜ Tools | 11.00Standard Mode |
12.50Thinking Level · High | 2.10Standard Mode | |
IMO 2024 数学推理 | 11.00Thinking Enabled | 19.00Thinking Enabled |
IMO 2025 数学推理 | 29.00Thinking Enabled | 15.20Thinking Enabled |
IMO-ProofBench 数学推理 | 59.00Thinking Enabled | 55.20Thinking Enabled |
20.00Thinking Enabled | 17.60Thinking Enabled | |
Terminal-Bench AI Agent - 工具使用 | 43.80Thinking Enabled ｜ Tools | 25.30Thinking Enabled |
MMMU 多模态理解 | 84.20Thinking Level · High | 82.00Thinking Enabled |
Simple Bench 常识推理 | 56.70Thinking Level · High | 62.40Thinking Enabled |
τ²-Bench - Telecom Agent能力评测 | 96.70Thinking Level · High ｜ Tools | 54.00Thinking Enabled ｜ Tools |
IF Bench 指令跟随 | 73.10Thinking Level · High | 49.00Thinking Enabled ｜ Tools |
BrowseComp AI Agent - 信息收集 | 54.90Thinking Enabled ｜ Tools | 7.80Thinking Enabled ｜ Tools |
Side-by-side input/output token pricing
Licensing, MoE architecture, and multi-modality support.
| Features & specs | GPT-5OpenAI | Gemini 2.5-ProGoogle Deep Mind |
|---|---|---|
Core specsRelease | 2025-08-07 | 2025-06-05 |
Context length | 400K | 1000K |
Max output | 131072 | 65536 |
MoE | No | No |
Supported modes | 常规模式（Non-Thinking Mode）思考模式（Thinking Mode）深度思考（Deeper Thinking Mode） | 常规模式（Non-Thinking Mode）思考模式（Thinking Mode）深度思考（Deeper Thinking Mode） |
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 | Introducing GPT-5 | Try the latest Gemini 2.5 Pro before general availability. |
DataLearner blog | OpenAI发布GPT-5：这是一个包含实时路由的AI系统，而不仅仅是一个模型 | Google发布Gemini 2.5 Pro: Gemini系列第一个2.5版本的模型，最高支持200万上下文，全模态输入，推理大模型，LMArena排名第一 |
Gemini 2.5-Pro
Google Deep Mind
