gpt-oss-120b
OpenAI · Released 5 Aug 2025 · gpt-oss-120b
–
Input price / 1M tokens
–
Output price / 1M tokens
131K
Context window
#68 of 71
164
Output tokens / second
#17 of 61
62.4%
SWE-bench Verified
#9 of 12
80.1%
GPQA Diamond
#23 of 27
97.9%
AIME 2025
#4 of 7
14.9%
Humanity’s Last Exam (no tools)
#19 of 23
Summary
gpt-oss-120b is an open-weight model from OpenAI, released on 5 Aug 2025.
Its 131K-token context window ranks #68 of 71.
Measured output speed is 164 tokens per second, #17 of 61.
OpenAI's larger open-weight reasoning model (117B total, 5.1B active parameters, MoE). OpenAI says it is near o4-mini on core reasoning benchmarks.
Benchmarks · 5 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| SWE-bench Verifiedhigh reasoning effort | 62.4% | #9 of 12 | OpenAI gpt-oss model cardSelf-reported | |
| GPQA Diamondno tools, high reasoning effort | 80.1% | #23 of 27 | OpenAI gpt-oss model cardSelf-reported | |
| AIME 2025with tools, high reasoning effort | 97.9% | #4 of 7 | OpenAI gpt-oss model cardSelf-reported | |
| Humanity’s Last Exam (no tools)no tools, high reasoning effort | 14.9% | #19 of 23 | OpenAI gpt-oss model cardSelf-reported | |
| MMLUhigh reasoning effort | 90% | – | OpenAI gpt-oss model cardSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with gpt-oss-120b, and the closest scores.
SWE-bench Verified
#9 of 12Real GitHub issues, fixed and tested
- gpt-oss-120b62.4
- gpt-oss-20b60.7
- Gemini 2.5 Pro59.6
- Gemini 2.5 Flash60.4
- Gemini 3 Flash78
- Gemini 3.1 Pro80.6
GPQA Diamond
#23 of 27Graduate-level science questions
- gpt-oss-120b80.1
- gpt-oss-20b71.5
- Gemini 2.5 Pro86.4
- Gemini 2.5 Flash82.8
- Gemini 3 Flash90.4
- Gemini 3.1 Pro94.3
AIME 2025
#4 of 7Competition mathematics
Humanity’s Last Exam (no tools)
#19 of 23Expert questions, no search or code
- gpt-oss-120b14.9
- gpt-oss-20b10.9
- Gemini 2.5 Pro21.6
- Gemini 2.5 Flash11
- Gemini 3 Flash33.7
- Gemini 3.1 Pro44.4
Compare gpt-oss-120b side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- context, max output, knowledge cutoff, parameters, license: developers.openai.com/api/docs/models/gpt-oss-120b, checked 2 Oct 2026
- release date Aug 5 2025, Apache 2.0, 128k context, architecture: openai.com/index/introducing-gpt-oss/, checked 2 Oct 2026
- benchmarks, license wording: arxiv.org/html/2508.10925v1, checked 2 Oct 2026
- absence from OpenAI API price list: developers.openai.com/api/docs/pricing, checked 2 Oct 2026
- output speed: artificialanalysis.ai/models/gpt-oss-120b, checked 2 Oct 2026
Building on gpt-oss-120b?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.