Individual experiment result
MMLU-ProX (DE)
GPT-5.5 OpenAI
Exact match reasoning on / locked provider-internal
87.2%
Exact match
Run details
- Date
- 2026-07-01
- Test cases
- 11,759
- Median latency
- 29.48s
- Total cost
- $193.51
- Avg. prompt tokens
- 1,594.7
- Avg. answer tokens
- 145.5
- Avg. reasoning tokens
- 137.3
- 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 |
|---|---|
| biology | 92.3% |
| business | 91.1% |
| chemistry | 89.8% |
| computer science | 90.0% |
| economics | 89.2% |
| engineering | 83.3% |
| health | 81.8% |
| history | 77.7% |
| law | 74.8% |
| math | 94.6% |
| other | 84.3% |
| philosophy | 85.2% |
| physics | 90.8% |
| psychology | 86.3% |
Provenance
- Canonical model
- openai/gpt-5.5
- Run ID
2026-07-01T09-02-53__openai__codex__gpt-5.5-low__mmlu_prox_de