Individual experiment result
MMLU-ProX (DE)
DeepSeek V4 Flash DeepSeek
Exact match reasoning off provider-internal
74.9%
Exact match
Run details
- Date
- 2026-06-16
- Test cases
- 11,759
- Median latency
- 1.98s
- Total cost
- $1.04
- Avg. prompt tokens
- 1,718.7
- Avg. answer tokens
- 162.2
- 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 | 88.4% |
| business | 76.2% |
| chemistry | 73.3% |
| computer science | 85.9% |
| economics | 75.0% |
| engineering | 64.1% |
| health | 71.3% |
| history | 68.2% |
| law | 46.1% |
| math | 88.7% |
| other | 69.4% |
| philosophy | 71.9% |
| physics | 84.7% |
| psychology | 80.8% |
Provenance
- Canonical model
- deepseek/deepseek-v4-flash
- Run ID
2026-06-16T12-31-34__openrouter__deepseek__deepseek-v4-flash__mmlu_prox_de