Individual experiment result
MuSR (DE)
Gemini 3.6 Flash Google
Accuracy reasoning off provider-internal
80.3%
Accuracy
Run details
- Date
- 2026-07-23
- Test cases
- 564
- Median latency
- 4.01s
- Total cost
- $4.05
- Avg. prompt tokens
- 1,464.9
- Avg. answer tokens
- 663.7
- Avg. reasoning tokens
- —
- 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 | 79.6% |
| object placements | 73.4% |
| team allocation | 82.8% |
Provenance
- Canonical model
- google/gemini-3.6-flash
- Run ID
2026-07-22T22-59-41__vertex_ai__gemini-3.6-flash__musr_de