GPT-5.5vsGPT-5.4
在 21 个共同 benchmark 中，GPT-5.5 整体领先：GPT-5.5 领先 19 项，GPT-5.4 领先 2 项，持平 0 项，平均分差 +6.73。
GPT-5.519 项(90%)(10%)2 项GPT-5.4
评测分数
按能力类目分组，每组内按分差大小排列；共 21 项。
Math and Reasoning
GPT-5.5 领先 5/5| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| MathArena Apex | 80.212 / 17Extra-High (With Tools) | 54.178 / 17Extra-High (With Tools) | +26.04 |
| FrontierMath Tier 4 v2 | 72.507 / 42Extra-High (No Tools) | 4912 / 42Extra-High (No Tools) | +23.50 |
| FrontierMath v2 | 85.266 / 58Extra-High (No Tools) | 78.6010 / 58Extra-High (No Tools) | +6.67 |
| AIME 2026 | 1001 / 29Extra-High (With Tools) | 99.174 / 29Extra-High (With Tools) | +0.83 |
| HMMT Feb 2026 | 98.481 / 10Extra-High (With Tools) | 97.732 / 10Extra-High (With Tools) | +0.75 |
Coding and Software Engineer
GPT-5.5 领先 3/3| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| DeepSWE | 67.0429 / 85Extra-High (With Tools) | 51.7758 / 85Extra-High (With Tools) | +15.27 |
| WeirdML v2 | 67.1522 / 52Normal (With Tools) | 57.4431 / 52Normal (With Tools) | +9.71 |
| GSO | 40.204 / 21Extra-High (With Tools) | 31.376 / 21Extra-High (With Tools) | +8.83 |
General Knowledge
GPT-5.5 领先 3/3| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| HLE | 13.70351 / 563Normal (No Tools) · Text only | 11.30374 / 563Normal (No Tools) · Text only | +2.40 |
| LiveBench | 79.911 / 117Deep Thinking (No Tools) | 77.975 / 117Deep Thinking (No Tools) | +1.94 |
| CritPt | 1.40135 / 200Normal (No Tools) | 0.60168 / 200Normal (No Tools) | +0.80 |
Agent Level Benchmark
GPT-5.5 领先 2/2| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| Terminal Bench Hard | 49.2024 / 244Normal (With Tools) | 37.9059 / 244Normal (With Tools) | +11.30 |
| τ²-Bench - Telecom | 69.30139 / 264Normal (With Tools) | 64.30151 / 264Normal (With Tools) | +5 |
AI Agent - Tool Usage
GPT-5.5 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| MCP-Atlas | 75.3025 / 43Extra-High (With Tools) | 70.6028 / 43Extra-High (With Tools) | +4.70 |
Claw-style Agent Evaluation
GPT-5.4 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| PinchBench v2 | 75.5018 / 45Reported best (effort unspecified) | 75.7017 / 45Reported best (effort unspecified) | -0.20 |
Instruction Following
GPT-5.4 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| IF Bench | 46.10174 / 282Normal (No Tools) | 48.40164 / 282Normal (No Tools) | -2.30 |
Long Context
GPT-5.5 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| AA-LCR | 64122 / 170Normal (No Tools) | 58.30130 / 170Normal (No Tools) | +5.70 |
Multimodal Understanding
GPT-5.5 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| MMMU-Pro | 71.40119 / 227Normal (No Tools) | 70.60123 / 227Normal (No Tools) | +0.80 |
Text Embedding
GPT-5.5 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| Context Arena | 41.9398 / 126Normal (No Tools) | 32.79112 / 126Normal (No Tools) | +9.14 |
Writing and Creative Capabilities
GPT-5.5 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| Creative Writing | 1,84412 / 106Normal (No Tools) | 1,83614 / 106Normal (No Tools) | +7.90 |
科学与综合推理
GPT-5.5 领先 1/1| 评测项 | GPT-5.5 | GPT-5.4 | 分差 |
|---|---|---|---|
| GPQA Diamond | 77.27264 / 462Normal (No Tools) | 74.80284 / 462Normal (No Tools) | +2.47 |
规格对比
| 字段 | GPT-5.5 | GPT-5.4 |
|---|---|---|
| 发布机构 | OpenAI | OpenAI |
| 发布时间 | 2026-04-23 | 2026-03-05 |
| 模型类型 | 推理大模型 | 多模态大模型 |
| 架构 | 稠密模型 | 稠密模型 |
| 参数规模 | 暂无数据 | 暂无数据 |
| 上下文长度 | 1000K | 1M |
| 最大输出 | 128K | 125K |
API 调用价格
价格优先使用 DataLearner 配置的 API 记录；缺失项不做推测。
| 价格项 | GPT-5.5 | GPT-5.4 |
|---|---|---|
| 文本输入 | $5 / 1M tokens | $2.5 / 1M tokens |
| 文本输出 | $30 / 1M tokens | $15 / 1M tokens |
| 缓存读取 | $0.5 / 1M tokens | $0.25 / 1M tokens |
| 缓存写入 | $6.25 / 1M tokens | 暂无公开价格 |
小结
- GPT-5.5在以下类目领先:Math and Reasoning (5/5)、Coding and Software Engineer (3/3)、General Knowledge (3/3)、Agent Level Benchmark (2/2)、AI Agent - Tool Usage (1/1)、Long Context (1/1)、Multimodal Understanding (1/1)、Text Embedding (1/1)、Writing and Creative Capabilities (1/1)、科学与综合推理 (1/1)
- GPT-5.4在以下类目领先:Claw-style Agent Evaluation (1/1)、Instruction Following (1/1)
21 个共同 benchmark 上，GPT-5.5 平均高出 6.73 分。
单项差距最大的 benchmark：MathArena Apex — GPT-5.5 80.21，GPT-5.4 54.17（分差 +26.04）。
本页正文由结构化模型、价格与 benchmark 数据生成，不使用实时 LLM 撰写。