by OpenAIRank #165Score 68.3
gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for lower-latency inference and deployability on consumer or single-GPU hardware. The model is trained in OpenAI’s Harmony response format and supports reasoning level configuration, fine-tuning, and agentic capabilities including function calling, tool use, and structured outputs.
| 信号 | 标准化 | 权重 | 贡献 | 新鲜度 |
|---|---|---|---|---|
Capabilities capability | 66.7 | 20% | 13.3 | 2026-03-26T07:22:37.352Z |
Benchmarks benchmark | 55.7 | 30% | 16.7 | 2026-03-26T07:22:37.352Z |
Pricing pricing_tier | 0.1 | 15% | 0.0 | 2026-03-26T07:22:37.352Z |
Context Window context_window | 81.2 | 10% | 8.1 | 2026-03-26T07:22:37.352Z |
Recency recency | 90.4 | 15% | 13.6 | 2026-03-26T07:22:37.352Z |
Output Capacity output_capacity | 85.3 | 10% | 8.5 | 2026-03-26T07:22:37.352Z |
| 功能 | 支持 |
|---|---|
| 视觉 | 否 |
| 推理 | 是 |
| JSON模式 | 是 |
| 流式输出 | 是 |
| 函数调用 | 是 |
| 网页搜索 | 否 |