GPT-6 AstravsGPT-5.6 Sol
在 14 个共同 benchmark 中，GPT-6 Astra 整体领先：GPT-6 Astra 领先 13 项，GPT-5.6 Sol 领先 1 项，持平 0 项，平均分差 +13.21。
GPT-6 Astra
OpenAI · 2026-09-03 · 推理大模型
GPT-5.6 Sol
OpenAI · 2026-06-26 · 推理大模型
GPT-6 Astra13 项(93%)(7%)1 项GPT-5.6 Sol
评测分数
按能力类目分组，每组内按分差大小排列；共 14 项。
General Knowledge
GPT-6 Astra 领先 5/5| 评测项 | GPT-6 Astra | GPT-5.6 Sol | 分差 |
|---|---|---|---|
| ARC-AGI-3 | 62.701 / 16Max (No Tools) | 7.803 / 16Max (No Tools) | +54.90 |
| AA Intelligence Index | 61.207 / 28Max (No Tools) | 4224 / 28Normal (With Tools) | +19.20 |
| HLE | 57.2012 / 190Max (With Tools) | 49.5039 / 190Max (No Tools) | +7.70 |
| ARC-AGI-2 | 951 / 85Max (No Tools) | 92.502 / 85Max (No Tools) | +2.50 |
| ARC-AGI-1 | 98.501 / 91Max (No Tools) | 97.504 / 91Extra-High (No Tools) | +1 |
AI Agent - Tool Usage
GPT-6 Astra 领先 3/3| 评测项 | GPT-6 Astra | GPT-5.6 Sol | 分差 |
|---|---|---|---|
| Terminal-Bench-Science 0.1 | 64.601 / 11Max (With Tools) | 22.404 / 11Max (With Tools) | +42.20 |
| Terminal-Bench 4.0 | 57.901 / 13Max (With Tools) | 37.276 / 13Max (With Tools) | +20.63 |
| OSWorld 2.0 | 72.602 / 9Max (With Tools) | 62.605 / 9Extra-High (With Tools) | +10 |
Coding and Software Engineer
GPT-6 Astra 领先 2/3| 评测项 | GPT-6 Astra | GPT-5.6 Sol | 分差 |
|---|---|---|---|
| AA Coding Agent Index | 674 / 4Max (With Tools) | 801 / 4Extra-High (With Tools) | -13 |
| FrontierCode 1.1 | 64.501 / 6Max (With Tools) | 60.604 / 6Max (With Tools) | +3.90 |
| DeepSWE | 74.102 / 35Max (With Tools) | 72.704 / 35Extra-High (With Tools) | +1.40 |
Agent Level Benchmark
GPT-6 Astra 领先 1/1| 评测项 | GPT-6 Astra | GPT-5.6 Sol | 分差 |
|---|---|---|---|
| Agents' Last Exam | 59.301 / 15Max (With Tools) | 52.702 / 15Extra-High (With Tools) | +6.60 |
Math and Reasoning
GPT-6 Astra 领先 1/1| 评测项 | GPT-6 Astra | GPT-5.6 Sol | 分差 |
|---|---|---|---|
| FrontierMath Tier 4 v2 | 97.601 / 41Max (No Tools) | 82.933 / 41Max (No Tools) | +14.67 |
科学与综合推理
GPT-6 Astra 领先 1/1| 评测项 | GPT-6 Astra | GPT-5.6 Sol | 分差 |
|---|---|---|---|
| GPQA Diamond | 961 / 271Max (No Tools) | 82.83122 / 271Normal (No Tools) | +13.17 |
规格对比
| 字段 | GPT-6 Astra | GPT-5.6 Sol |
|---|---|---|
| 发布机构 | OpenAI | OpenAI |
| 发布时间 | 2026-09-03 | 2026-06-26 |
| 模型类型 | 推理大模型 | 推理大模型 |
| 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 | 1.05M | 1.05M |
| 最大输出 | 128K | 128K |
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-6 Astra | GPT-5.6 Sol |
|---|---|---|
| 文本输入 | $10 / 1M tokens | $4 / 1M tokens |
| 文本输出 | $50 / 1M tokens | $20 / 1M tokens |
| 缓存读取 | $1 / 1M tokens | $0.4 / 1M tokens |
| 缓存写入 | $12.5 / 1M tokens | $5 / 1M tokens |
小结
- GPT-6 Astra在以下类目领先:General Knowledge (5/5)、AI Agent - Tool Usage (3/3)、Coding and Software Engineer (2/3)、Agent Level Benchmark (1/1)、Math and Reasoning (1/1)、科学与综合推理 (1/1)
14 个共同 benchmark 上，GPT-6 Astra 平均高出 13.21 分。
单项差距最大的 benchmark：ARC-AGI-3 — GPT-6 Astra 62.70，GPT-5.6 Sol 7.80（分差 +54.90）。
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。