Individual experiment result
MuSR (DE)
Gemini 3.8 Flash Google
Accuracy reasoning on / locked provider-internal
89.0%
Accuracy
Run details
- Date
- 2026-09-03
- Test cases
- 553
- Median latency
- 20.21s
- Total cost
- —
- Avg. prompt tokens
- 1,484.9
- Avg. answer tokens
- 462.7
- Avg. reasoning tokens
- 1,642.2
- Quantization
- provider-internal
About this benchmark
German MuSR v1.1 defect-filtered benchmark: 553 questions (11 truncated team-allocation translations excluded), generate_until cot+; primary metric accuracy.
Subject breakdown
| Subject | Score |
|---|---|
| murder mystery | 91.6% |
| object placements | 75.0% |
| team allocation | 90.0% |
Provenance
- Canonical model
- google/gemini-3.8-flash
- Run ID
2026-09-03T14-49-11__vertex_ai__gemini-3.8-flash__musr_de