OpenAI
gpt-oss-20B (low) is part of OpenAI's gpt-oss open-weight model family, intended for developers who need strong local or self-hosted language-model capability. The local profile is useful for checking how this deployment-friendly model compares with hosted frontier systems on quality, throughput, and pricing assumptions.
Introducing gpt-ossRank #341 across 526
Rank #271 across 436
Rank #22 across 357
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Higher bars mean stronger relative placement.
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 14.3 | #341 |
| Artificial Analysis Coding Index | coding | 14.4 | #271 |
| Artificial Analysis Math Index | math | 62.3 | #113 |
| MMLU-Pro | reasoning | 71.8% | #208 |
| reasoning |
| 61.1% |
| #289 |
| Humanity's Last Exam | reasoning | 5.1% | #306 |
| LiveCodeBench | coding | 65.2% | #96 |
| SciCode | coding, reasoning | 34.0% | #222 |
| Output Speed | speed | 224.2 tok/s | #22 |
| Time to First Token | speed | 0.48s | #29 |
| Blended Price | cost | $0.095/M | #22 |
| Input Price | cost | $0.060/M | #27 |
| Output Price | cost | $0.200/M | #27 |
| Value Index | cost, overall | 150.5 | #32 |