openai/gpt-oss-120b
gpt-oss-120b is an open-weight, 117B-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...
| Provider | In $/M | Out $/M | Context | Uptime |
|---|---|---|---|---|
| DekaLLMbf16 | $0.039 | $0.180 | 131K | 99.5% |
| DeepInfrabf16 | $0.039 | $0.190 | 131K | 100% |
| WandBfp4 | $0.040 | $0.140 | 131K | 100% |
| Novitafp4 | $0.050 | $0.250 | 131K | 100% |
| SiliconFlowfp8 | $0.050 | $0.450 | 131K | 92.9% |
| DigitalOcean | $0.075 | $0.525 | 128K | 99.8% |
| $0.090 | $0.360 | 131K | 99.8% | |
| BaseTenfp4 | $0.100 | $0.500 | 128K | 100% |
| Parasailfp4 | $0.100 | $0.750 | 131K | 100% |
| SambaNova | $0.140 | $0.950 | 131K | 95.2% |
| Amazon Bedrock | $0.150 | $0.600 | 131K | 100% |
| Phala | $0.150 | $0.600 | 131K | 70.8% |
| Amazon Bedrock | $0.150 | $0.600 | 131K | — |
| Together | $0.150 | $0.600 | 131K | 97% |
| Groq | $0.150 | $0.600 | 131K | 100% |
| Nebiusfp4 | $0.150 | $0.600 | 131K | 100% |
| DeepInfrabf16 | $0.150 | $0.600 | 131K | 98% |
| Mara | $0.150 | $0.750 | 131K | 71.2% |
| Cerebrasfp16 | $0.350 | $0.750 | 131K | 100% |
gpt-oss-120b costs $0.039 per million input tokens and $0.180 per million output tokens via OpenRouter, making it 10th cheapest of 298 paid models.
gpt-oss-120b scores 33.3 on the Artificial Analysis Intelligence Index, ranking 66th of 178 benchmarked models, with a GPQA Diamond score of 78%.
gpt-oss-120b generates around 547 tokens per second with 177ms time-to-first-token (p50), the 4th fastest tracked model.
gpt-oss-120b supports a 131K-token context window. It accepts text input.