Individual experiment result
MMLU-ProX (DE)
Gemini 3.7 Flash Google
Exact match reasoning on / locked provider-internal
88.2%
Exact match
Run details
- Date
- 2026-08-14
- Test cases
- 11,759
- Median latency
- 4.60s
- Total cost
- $34.62
- Avg. prompt tokens
- 1,664.9
- Avg. answer tokens
- 191.5
- Avg. reasoning tokens
- 260.7
- 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 |
|---|---|
| biology | 94.0% |
| business | 90.6% |
| chemistry | 91.4% |
| computer science | 90.7% |
| economics | 89.9% |
| engineering | 84.5% |
| health | 82.1% |
| history | 80.1% |
| law | 76.6% |
| math | 95.4% |
| other | 83.8% |
| philosophy | 85.4% |
| physics | 93.0% |
| psychology | 87.5% |
Provenance
- Canonical model
- google/gemini-3.7-flash
- Run ID
2026-08-14T19-01-34__vertex_ai__gemini-3.7-flash__mmlu_prox_de