OpenAI's most powerful reasoning model. o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks. It also excels at technical writing and instruction-following. Use it to think through multi-step problems that involve analysis across text, code, and images.
发布日期2025年4月16日
参数规模—
上下文长度200K
许可证Proprietary
知识截止2024年5月31日
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| COLLIE | language reasoning writing | 98.4 | 来源 |
| 安全防御榜·Bias | security | 97.0 | — |
| 安全防御榜·Info Safety | security | 94.8 | — |
| 安全防御榜·Malicious Use | security | 93.2 | — |
| 安全防御榜 | security | 92.4 | — |
| 安全防御榜·Socioeconomic | security | 92.3 | — |
| 安全防御榜·Human Rights | security | 92.2 | — |
| 开源模型能力榜·IFEval | general | 92.2 | 来源 |
| AIME 2024 | math reasoning | 91.6 | 来源 |
| 安全防御榜·Toxicity | security | 91.2 | — |
| ARC-AGI | reasoning vision spatial_reasoning | 88.0 | 来源 |
| MathVista | math vision multimodal | 86.8 | 来源 |
| AIME 2025 | math reasoning | 86.4 | 来源 |
| 安全防御榜·Child Safety | security | 85.7 | — |
| 开源模型能力榜·AIME2025 | general | 85.1 | 来源 |
| GPQA | reasoning general | 83.3 | 来源 |
| VideoMMMU | multimodal vision reasoning | 83.3 | 来源 |
| 开源模型能力榜·MMLU-Pro | general | 83.0 | 来源 |
| MMMU | multimodal reasoning general | 82.9 | 来源 |
| 开源模型能力榜·GPQA-Diamond | general | 82.3 | 来源 |
| Aider-Polyglot | general code | 81.3 | 来源 |
| Tau2 retail | communication reasoning | 80.2 | 来源 |
| CharXiv-R | reasoning vision multimodal | 78.6 | 来源 |
| MMMU-Pro | vision multimodal reasoning general | 76.4 | 来源 |
| 开源模型能力榜 | general | 72.8 | 来源 |
| 开源模型能力榜·LiveCodeBenchV6 | general | 72.7 | 来源 |
| SWE-Bench Verified | reasoning frontend_development code | 69.1 | 来源 |
| Tau2 airline | reasoning communication | 64.8 | 来源 |
| ERQA | vision reasoning spatial_reasoning | 64.0 | 来源 |
| Tau-bench | general reasoning | 63.0 | 来源 |
| Scale MultiChallenge | reasoning communication general | 60.4 | 来源 |
| Tau2 telecom | communication reasoning | 58.2 | 来源 |
| Scale MultiChallenge | reasoning communication general | 56.5 | 来源 |
| BrowseComp | reasoning search | 49.7 | 来源 |
| Humanity's Last Exam | general | 24.3 | 来源 |
| 开源模型能力榜·HLE | general | 21.3 | 来源 |
| Humanity's Last Exam | general | 20.2 | 来源 |
| FrontierMath | math reasoning | 15.8 | 来源 |
| Humanity's Last Exam | general | 14.7 | 来源 |
| ARC-AGI v2 | reasoning vision spatial_reasoning | 6.5 | 来源 |
Pricing
API 价格对比
| 服务商 | 输入价 | 输出价 | 上下文 | 吞吐(tok/s) | 延迟(s) | 函数调用 | 代码执行 | 联网搜索 |
|---|---|---|---|---|---|---|---|---|
| Abacus | $2.00 | $8.00 | 200K | — | — | ✓ | ✗ | ✗ |
| Azure | $2.00 | $8.00 | 200K | — | — | ✓ | ✗ | ✗ |
| Azure Cognitive Services | $2.00 | $8.00 | 200K | — | — | ✓ | ✗ | ✗ |
| Helicone | $2.00 | $8.00 | 200K | — | — | ✓ | ✗ | ✗ |
| LLM Gateway | $2.00 | $8.00 | 200K | — | — | ✓ | ✗ | ✗ |
| OpenAI | $2.00 | $8.00 | 200K | — | — | ✓ | ✗ | ✗ |
| Jiekou.AI | $10.00 | $40.00 | 131K | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。