Individual experiment result
MMLU-ProX (DE)
Gemini 3.1 Pro Google
Exact match reasoning on / locked provider-internal
89.2%
Exact match
Run details
- Date
- 2026-08-15
- Test cases
- 11,759
- Median latency
- 3.46s
- Total cost
- $95.16
- Avg. prompt tokens
- 1,668.9
- Avg. answer tokens
- 243
- Avg. reasoning tokens
- 153.3
- Quantization
- provider-internal
About this benchmark
German MMLU-ProX across 14 subjects; primary metric accuracy. Legacy rows have 11,759 items. Corrected v1.1 rows have 11,737 after a frozen, model-independent manifest excludes 20 duplicated-gold translation collisions, one independently reviewed ambiguous source item, and one independently confirmed notation-translation collapse; the row's item count identifies the definition.
Subject breakdown
| Subject | Score |
|---|---|
| nothink biology | 95.1% |
| nothink business | 91.8% |
| nothink chemistry | 91.0% |
| nothink computer science | 91.5% |
| nothink economics | 91.4% |
| nothink engineering | 86.0% |
| nothink health | 82.8% |
| nothink history | 82.7% |
| nothink law | 78.6% |
| nothink math | 95.0% |
| nothink other | 85.1% |
| nothink philosophy | 86.0% |
| nothink physics | 93.6% |
| nothink psychology | 90.2% |
Provenance
- Canonical model
- google/gemini-3.1-pro-preview
- Run ID
2026-08-15T13-46-32__vertex_ai__gemini-3.1-pro-preview__mmlu_prox_de_nothink