reasoning benchmark
VoiceBench Avg
VoiceBench is the first benchmark designed to provide a multi-faceted evaluation of LLM-based voice assistants, evaluating capabilities including general knowledge, instruction-following, reasoning, and safety using both synthetic and real spoken instruction data with diverse speaker characteristics and environmental conditions.
Updated Aug 7, 2026
Published models1
Registry coverage1
MetricScore
EvidenceB
VoiceBench Avg leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AC | 74.1% | 100.0% | 1 | C |
VoiceBench Avg highlights
The top published results on this benchmark's own scale.
What is VoiceBench Avg?
Definition and scoring fields from the benchmark registry.
VoiceBench is the first benchmark designed to provide a multi-faceted evaluation of LLM-based voice assistants, evaluating capabilities including general knowledge, instruction-following, reasoning, and safety using both synthetic and real spoken instruction data with diverse speaker characteristics and environmental conditions.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- VoiceBench Avg
- Modality
- multimodal
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- voicebench-avg|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about VoiceBench Avg.
Which model scores highest on VoiceBench Avg?
Qwen2.5-Omni-7B is currently ranked first with 74.1%.
What does VoiceBench Avg measure?
VoiceBench is the first benchmark designed to provide a multi-faceted evaluation of LLM-based voice assistants, evaluating capabilities including general knowledge, instruction-following, reasoning, and safety using both synthetic and real spoken instruction data with diverse speaker characteristics and environmental conditions.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.