GPT-4.1 nano is OpenAI's fastest and cheapest model available in the GPT-4.1 family. It delivers exceptional performance at a small size with its 1 million token context window. Ideal for tasks like classification or autocompletion.
发布日期2025年4月14日
参数规模—
上下文长度1.0M
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 |
|---|---|---|---|
| 安全防御榜·Toxicity | security | 94.2 | — |
| 安全防御榜·Info Safety | security | 92.7 | — |
| 安全防御榜·Malicious Use | security | 90.6 | — |
| 安全防御榜 | security | 89.6 | — |
| 安全防御榜·Human Rights | security | 89.3 | — |
| 安全防御榜·Bias | security | 87.8 | — |
| 安全防御榜·Socioeconomic | security | 87.6 | — |
| 安全防御榜·Child Safety | security | 85.1 | — |
| MMLU | general reasoning language math | 80.1 | 来源 |
| IFEval | general | 74.5 | 来源 |
| CharXiv-D | reasoning vision multimodal | 73.9 | 来源 |
| 多模态开源模型榜·OCRBench | vision | 71.0 | 来源 |
| 多模态开源模型榜·AI2D | vision | 68.0 | 来源 |
| MMMLU | language reasoning math general | 66.9 | 来源 |
| 多模态开源模型榜·MMVet | vision | 62.8 | 来源 |
| 多模态开源模型榜·MMBench TEST | vision | 62.4 | 来源 |
| 多模态开源模型榜·MMMU VAL | vision | 57.6 | 来源 |
| 多模态开源模型榜 | vision | 57.5 | 来源 |
| Multi-IF | reasoning communication language | 57.2 | 来源 |
| MathVista | math vision multimodal | 56.2 | 来源 |
| MMMU | multimodal reasoning general | 55.4 | 来源 |
| 多模态开源模型榜·MathVista | vision | 53.7 | 来源 |
| 价值对齐榜·Fairness | security | 51.9 | — |
| GPQA | reasoning general | 50.3 | 来源 |
| 价值对齐榜·Legality | security | 48.0 | — |
| 多模态开源模型榜·MMStar | vision | 47.5 | 来源 |
| 价值对齐榜·Data Protection | security | 46.8 | — |
| 价值对齐榜 | security | 44.7 | — |
| COLLIE | language reasoning writing | 42.5 | 来源 |
| CharXiv-R | reasoning vision multimodal | 40.5 | 来源 |
| 价值对齐榜·Morality | security | 39.6 | — |
| 价值对齐榜·Safety | security | 37.2 | — |
| 多模态开源模型榜·HallusionBench | vision | 36.9 | 来源 |
| OpenAI-MRCR: 2 needle 128k | long_context reasoning | 36.6 | 来源 |
| Internal API instruction following (hard) | general | 31.6 | 来源 |
| MultiChallenge (o3-mini grader) | reasoning language | 31.1 | 来源 |
| AIME 2024 | math reasoning | 29.4 | 来源 |
| Graphwalks BFS <128k | reasoning spatial_reasoning | 25.0 | 来源 |
| TAU-bench Retail | reasoning communication | 22.6 | 来源 |
| MultiChallenge | communication reasoning | 15.0 | 来源 |
| TAU-bench Airline | reasoning communication | 14.0 | 来源 |
| OpenAI-MRCR: 2 needle 1M | long_context reasoning | 12.0 | 来源 |
| Aider-Polyglot | general code | 9.8 | 来源 |
| Graphwalks parents <128k | reasoning spatial_reasoning | 9.4 | 来源 |
| Aider-Polyglot Edit | general code | 6.2 | 来源 |
| ComplexFuncBench | long_context reasoning | 5.7 | 来源 |
| Graphwalks parents >128k | reasoning spatial_reasoning long_context | 5.6 | 来源 |
| Graphwalks BFS >128k | reasoning spatial_reasoning long_context | 2.9 | 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 |
|---|---|---|---|---|---|---|---|---|
| SAP AI Core | $0.08 | $0.26 | 1.0M | — | — | ✓ | ✗ | ✗ |
| 302.AI | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Abacus | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Azure | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Azure Cognitive Services | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Helicone | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| LLM Gateway | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| OpenAI | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Pioneer | $0.10 | $0.40 | 1.0M | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。