Individual experiment result
MuSR (DE)
Gemini 3 Flash Preview Google
Accuracy reasoning off provider-internal
79.3%
Accuracy
Run details
- Date
- 2026-06-25
- Test cases
- 564
- Median latency
- 5.41s
- Total cost
- $1.47
- Avg. prompt tokens
- 1,464.9
- Avg. answer tokens
- 622.9
- 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 | 78.8% |
| object placements | 76.6% |
| team allocation | 80.4% |
Provenance
- Canonical model
- google/gemini-3-flash-preview
- Run ID
2026-06-25T11-34-32__vertex_ai__gemini-3-flash-preview__musr_de