German Language LLM Index
a PeerBench project

Individual experiment result

MuSR (DE)

Tencent HY4-Preview Tencent

Accuracy reasoning off fp8
80.1%
Accuracy

Run details

Date
2026-08-30
Test cases
553
Median latency
16.08s
Total cost
Avg. prompt tokens
1,730.6
Avg. answer tokens
929.8
Avg. reasoning tokens
Quantization
fp8

About this benchmark

German MuSR v1.1 defect-filtered benchmark: 553 questions (11 truncated team-allocation translations excluded), generate_until cot+; primary metric accuracy.

2–5 option multiple choice translated · German translation Source ↗

Subject breakdown

Subject Score
murder mystery 78.0%
object placements 79.7%
team allocation 82.4%

Provenance

Canonical model
tencent/hy4-preview
Run ID
2026-08-30T19-30-04__Tencent__tencent__hy4-preview__musr_de