GPT-OSS-120B is an open-weight, 116.8B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized to run on a single H100 GPU with native MXFP4 quantization. The model supports configurable reasoning depth, full chain-of-thought access, and native tool use, including function calling, browsing, and structured output generation. It achieves near-parity with OpenAI o4-mini on core reasoning benchmarks. Note: While referred to as '120b' for simplicity, it technically has 116.8B parameters.
发布日期2025年8月5日
参数规模116.8B
上下文长度131K
许可证Apache 2.0
知识截止—
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| 开源模型能力榜·AIME2025 | general | 93.4 | 来源 |
| 开源模型能力榜·IFEval | general | 90.2 | 来源 |
| MMLU benchmark | general reasoning language math | 90.0 | 来源 |
| Codeforces Competition code | math reasoning | 87.4 | 来源 |
| Codeforces Competition code | math reasoning | 82.1 | 来源 |
| GPQA | reasoning general | 80.1 | 来源 |
| 开源模型能力榜·MMLU-Pro | general | 79.7 | 来源 |
| 开源模型能力榜·GPQA-Diamond | general | 78.9 | 来源 |
| 开源模型能力榜·LiveCodeBenchV6 | general | 78.4 | 来源 |
| 开源模型能力榜 | general | 73.2 | 来源 |
| TAU-bench Retail benchmark | reasoning communication | 67.8 | 来源 |
| HealthBench - Realistic health conversations | healthcare | 57.6 | 来源 |
| HealthBench Hard - Challenging health conversations | healthcare | 30.0 | 来源 |
| Humanity's Last Exam | general | 19.0 | 来源 |
| 开源模型能力榜·HLE | general | 18.3 | 来源 |
| Humanity's Last Exam | general | 14.9 | 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 |
|---|---|---|---|---|---|---|---|---|
| Cortecs | $0.00 | $0.00 | 128K | — | — | ✓ | ✗ | ✗ |
| Kenari | $0.00 | $0.00 | 131K | — | — | ✓ | ✗ | ✗ |
| Privatemode AI | $0.00 | $0.00 | 128K | — | — | ✓ | ✗ | ✗ |
| Helicone | $0.04 | $0.16 | 131K | — | — | ✓ | ✗ | ✗ |
| LLM Gateway | $0.05 | $0.25 | 131K | — | — | ✓ | ✗ | ✗ |
| DInference | $0.07 | $0.27 | 131K | — | — | ✓ | ✗ | ✗ |
| Neon | $0.07 | $0.28 | 131K | — | — | ✓ | ✗ | ✗ |
| OVHcloud AI Endpoints | $0.09 | $0.47 | 131K | — | — | ✓ | ✗ | ✗ |
| AKI.IO | $0.15 | $0.55 | 128K | — | — | ✓ | ✗ | ✗ |
| FrogBot | $0.15 | $0.60 | 131K | — | — | ✓ | ✗ | ✗ |
| Scaleway | $0.15 | $0.60 | 128K | — | — | ✓ | ✗ | ✗ |
| Tinfoil | $0.15 | $0.60 | 131K | — | — | ✓ | ✗ | ✗ |
| Cerebras | $0.35 | $0.75 | 131K | — | — | ✓ | ✗ | ✗ |
| Regolo AI | $1.00 | $4.20 | 128K | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。