Individual experiment result
MMLU-ProX (DE)
Qwen3.6 27B (BF16) Alibaba
Exact match reasoning off bf16
82.3%
Exact match
Run details
- Date
- 2026-08-15
- Test cases
- 11,737
- Median latency
- 18.46s
- Total cost
- —
- Avg. prompt tokens
- 1,689.5
- Avg. answer tokens
- 932.8
- Avg. reasoning tokens
- —
- Quantization
- bf16
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 | 89.8% |
| business | 87.7% |
| chemistry | 90.2% |
| computer science | 88.5% |
| economics | 86.2% |
| engineering | 77.3% |
| health | 75.1% |
| history | 72.6% |
| law | 57.1% |
| math | 93.7% |
| other | 74.3% |
| philosophy | 73.9% |
| physics | 89.7% |
| psychology | 82.5% |
Provenance
- Canonical model
- qwen/qwen3.6-27b
- Run ID
2026-08-15T17-46-46__openai__Qwen__Qwen3.6-27B__mmlu_prox_de