Skip to content

Benchmark

WenetSpeech · Meeting WER

Scores are percentages. Lower is better.

Models and systems measured
6
Scores recorded
7
Sources
1
IQ weight
0.00

Score distribution

12.1%
Mean
5.5%
Median
13.1%
Std dev
4.7%
10th percentile
16.7%
90th percentile
4.6%
Min
4.6%42.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.00 for chat models, measured over 5 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1Qwen3.8 Omni FlashQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.6%
2Qwen3.5 Omni PlusQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.8%
3Qwen3.5 Omni FlashQwen3.5-Omni release · ASR WER/CER · lower is better · Meeting split· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.5.5%
4Qwen3.5 Omni PlusQwen3.5-Omni release · ASR WER/CER · lower is better · Meeting split· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.5.8%
5Gemini 3.8 FlashQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.16.7%
6Muse Spark 1.2Qwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.42.6%

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Seed-2.0-LiteQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.7%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.