← Overview
gpt-oss-120b
- Quantization
- MXFP4
- Reasoning
- on
- Profile
- quick
- Suite
- v2b
- Date
- 2026-09-18 – 2026-09-19
- Hardware
- NVIDIA DGX Spark, 128 GB RAM
- Inference software
- vLLM 0.26.0
- Measured
- directly at the inference server
- Overall
- 73.1
Categories
| Category | Score |
|---|
| Knowledge & reasoning | 68.5 |
| Math | 70.4 |
| Instruction following | 66.1 |
| Code | 83.0 |
| German | 77.4 |
Tasks
| Category | Task | Score | ± | n |
|---|
| Knowledge & reasoning |
mmlu_pro |
71.0 |
± 2.6 |
300 |
| Knowledge & reasoning |
gpqa_diamond_gen |
68.2 |
± 3.3 |
198 |
| Knowledge & reasoning |
musr_gen |
66.4 |
± 3.0 |
250 |
| Math |
omni_math_gen |
70.4 |
± 2.9 |
250 |
| Instruction following |
ifbench |
60.0 |
± 2.8 |
300 |
| Instruction following |
ifeval |
84.7 |
± 3.0 |
150 |
| Code |
livecodebench_gen |
83.0 |
± 2.7 |
200 |
| German |
mmlu_prox_de |
80.3 |
± 2.3 |
300 |
| German |
include_de_gen |
71.4 |
± 3.7 |
154 |
± is the standard error; the 95% interval spans roughly 1.96 standard errors around the score.
Speed
| Measure | Context (target) | Context (actual) | Median | p10 | p90 | n |
|---|
| No speed measurement. |
Measured directly at the inference server.