MU-Bench
Gemini 3.5 Transcribe 14.9% leads the models we track Speech-to-text providers transcribe 4,270 caller utterances from real 8 kHz customer service phone calls in five locales, scored against human word-level transcripts.
Locales are en-US, es-MX, tr-TR, vi-VN and zh-CN. Data is CC BY-NC 4.0.
- Made by
- Sierra
- Released
- April 2026
- Kind
- Speech recognition and translation
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Last model added
- August 2026
- Scores refreshed
- November 12th, 2026
Chirp-3 tops the leaderboard with 14.1%
Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Gemini 3.5 Transcribe, at 14.9%.
The newest model with a score here came out in August 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- Claude Fable 5.2
- Gemini 4 Carbon
- GPT-6 Astra
- Claude Opus 5.5
- and 16 more
- 1 Chirp-3 batch; us 14.1%
- 2 Gemini 3.5 Transcribe batch; us 14.9%
- 3 Scribe-v2 batch; us-east-2 21.2%
- 4
Qwen3-ASR-1.7B batch; us 23.6% - 5 Azure-Speech-v1 batch; us-east-2 24.6%
- 6 gpt-4o-mini-transcribe batch; us-east-2 27.3%
- 7 Grok-STT batch; us-east-1 32.1%
- 8 Nova-3 batch; us-east-2 32.6%
Source: Official MU-Bench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Chirp-3 | 14.1% | batch; us, Google | Apr 2026 | Official leaderboard |
| 2 | Gemini 3.5 Transcribe | 14.9% | batch; us, Google | Aug 2026 | Official leaderboard |
| 3 | Scribe-v2 | 21.2% | batch; us-east-2, ElevenLabs | Apr 2026 | Official leaderboard |
| 4 | Qwen3-ASR-1.7B | 23.6% | batch; us, Baseten | Jan 2026 | Official leaderboard |
| 5 | Azure-Speech-v1 | 24.6% | batch; us-east-2, Microsoft | Apr 2026 | Official leaderboard |
| 6 | gpt-4o-mini-transcribe | 27.3% | batch; us-east-2, OpenAI | Mar 2025 | Official leaderboard |
| 7 | Grok-STT | 32.1% | batch; us-east-1, xAI | May 2026 | Official leaderboard |
| 8 | Nova-3 | 32.6% | batch; us-east-2, Deepgram | Apr 2026 | Official leaderboard |
2 models have held number one on the official leaderboard since its release in April 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official MU-Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official MU-Bench leaderboard
No lab model card reports it yet
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.
More in Audio and video
- Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
- LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
- MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
- MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
- FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better.
Phi-4-multimodal 4.00% - VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%