Individual experiment result
MMLU-ProX (DE)
Gemini 3.5 Flash Google
Exact match reasoning off provider-internal
86.5%
Exact match
Run details
- Date
- 2026-05-26
- Test cases
- 11,759
- Median latency
- 2.23s
- Total cost
- $66.76
- Avg. prompt tokens
- 1,664.9
- Avg. answer tokens
- 353.3
- Avg. reasoning tokens
- —
- 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 | 92.9% |
| business | 89.6% |
| chemistry | 89.1% |
| computer science | 87.6% |
| economics | 89.1% |
| engineering | 82.5% |
| health | 78.6% |
| history | 80.6% |
| law | 72.8% |
| math | 94.6% |
| other | 82.4% |
| philosophy | 83.2% |
| physics | 90.4% |
| psychology | 87.8% |
Provenance
- Canonical model
- google/gemini-3.5-flash
- Run ID
2026-05-26T15-52-24__vertex_ai__gemini-3.5-flash