ScreenASR-Bench
Models transcribe speech using on-screen text, visible objects, or clues from the scene. The authors report word error rate and recall.
- Made by
- Mingwei Fu and colleagues
- Released
- September 2026
- Kind
- Speech recognition and translation
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision, Audio and speech
- Last model added
- April 2026
- Scores refreshed
- November 20th, 2026
Qwen3.5-Omni-Plus leads with 3.47% among models released by April 2026 Qwen3.5-Omni-Plus leads with 3.47% among models released by April 2026
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in April 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1
Qwen3.5-Omni-Plus Audio + Caption 3.47%
- 2
Doubao Seed 2.0 Lite Apr 2026, Audio + Caption 3.94%
- 3 Seed-ASR-2.0 Audio + Caption 3.96%
- 4 Qwen3-Omni Audio + Caption 4.01%
- 5 MiMo-V2.5 Audio + Caption 4.83%
- 6 MOSS-Audio-8B-Instruct Audio + Caption 5.15%
- 7 MiMo-V2.5-ASR Audio only 5.77%
- 8 Qwen-Audio-3.0-ASR Audio + Caption 5.80%
- 9
Qwen3-ASR-1.7B Audio + Caption 5.81%
- 10 FireRedASR2-LLM Audio only 6.22%
- 11 MiMo-Audio-7B-Instruct Audio + Caption 6.52%
- 12 StepAudio-2.5-Chat Audio only 7.94%
Source: The ScreenASR-Bench paper
- 1
Qwen3.5-Omni-Plus Audio + Caption 3.47%
- 2
Doubao Seed 2.0 Lite Apr 2026, Audio + Caption 3.94%
- 3 MiMo-V2.5 Audio + Caption 4.83%
- 4
Qwen3-ASR-1.7B Audio + Caption 5.81%
- 5 MiMo-Audio-7B-Instruct Audio + Caption 6.52%
Source: The ScreenASR-Bench paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 3.47% | Audio + Caption | Mar 2026 | Official leaderboard |
| 2 | | 3.94% | Audio + Caption | Apr 2026 | Official leaderboard |
| 3 | Seed-ASR-2.0 | 3.96% | Audio + Caption | Official leaderboard | |
| 4 | Qwen3-Omni | 4.01% | Audio + Caption | Official leaderboard | |
| 5 | MiMo-V2.5 | 4.83% | Audio + Caption | Apr 2026 | Official leaderboard |
| 6 | MOSS-Audio-8B-Instruct | 5.15% | Audio + Caption | Official leaderboard | |
| 7 | MiMo-V2.5-ASR | 5.77% | Audio only | Official leaderboard | |
| 8 | Qwen-Audio-3.0-ASR | 5.80% | Audio + Caption | Official leaderboard | |
| 9 | | 5.81% | Audio + Caption | Jan 2026 | Official leaderboard |
| 10 | FireRedASR2-LLM | 6.22% | Audio only | Official leaderboard | |
| 11 | MiMo-Audio-7B-Instruct | 6.52% | Audio + Caption | Sep 2025 | Official leaderboard |
| 12 | StepAudio-2.5-Chat | 7.94% | Audio only | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 3.47% | Audio + Caption | Mar 2026 | Official leaderboard |
| 2 | | 3.94% | Audio + Caption | Apr 2026 | Official leaderboard |
| 3 | MiMo-V2.5 | 4.83% | Audio + Caption | Apr 2026 | Official leaderboard |
| 4 | | 5.81% | Audio + Caption | Jan 2026 | Official leaderboard |
| 5 | MiMo-Audio-7B-Instruct | 6.52% | Audio + Caption | Sep 2025 | Official leaderboard |
Rankings by part
We read 6 parts from the benchmark’s paper, each ranking models on one part of the test, such as ETE WER, ETE recall and EVE WER. Qwen3.5-Omni-Plus leads 5 of them, and MiMo-V2.5 leads the other one. Arrows show how far a model moved from its overall rank among the models we track.
ETE WER
5 models we track have a score here.
- 1
Qwen3.5-Omni-Plus 3.68%
- 2
Doubao Seed 2.0 Lite Apr 2026 4.17%
- 3 MiMo-V2.5 5.35%
- 4
Qwen3-ASR-1.7B 6.08%
- 5 MiMo-Audio-7B-Instruct 6.90%
ETE recall
5 models we track have a score here.
- 1 MiMo-V2.5 ↑2 90.2%
- 2
Doubao Seed 2.0 Lite Apr 2026 85.7%
- 3
Qwen3.5-Omni-Plus ↓2 84.9%
- 4 MiMo-Audio-7B-Instruct ↑1 82.1%
- 5
Qwen3-ASR-1.7B ↓1 77.9%
EVE WER
5 models we track have a score here.
- 1
Qwen3.5-Omni-Plus 3.00%
- 2
Doubao Seed 2.0 Lite Apr 2026 3.50%
- 3 MiMo-V2.5 4.27%
- 4
Qwen3-ASR-1.7B 5.34%
- 5 MiMo-Audio-7B-Instruct 6.04%
EVE recall
5 models we track have a score here.
- 1
Qwen3.5-Omni-Plus 78.1%
- 2 MiMo-V2.5 ↑1 77%
- 3
Doubao Seed 2.0 Lite Apr 2026 ↓1 72.6%
- 4 MiMo-Audio-7B-Instruct ↑1 60.3%
- 5
Qwen3-ASR-1.7B ↓1 52.8%
ISE WER
5 models we track have a score here.
- 1
Qwen3.5-Omni-Plus 3.76%
- 2
Doubao Seed 2.0 Lite Apr 2026 4.19%
- 3 MiMo-V2.5 4.79%
- 4
Qwen3-ASR-1.7B 6.03%
- 5 MiMo-Audio-7B-Instruct 6.56%
ISE recall
5 models we track have a score here.
- 1
Qwen3.5-Omni-Plus 73.1%
- 2 MiMo-V2.5 ↑1 71.3%
- 3
Doubao Seed 2.0 Lite Apr 2026 ↓1 66.8%
- 4 MiMo-Audio-7B-Instruct ↑1 51.1%
- 5
Qwen3-ASR-1.7B ↓1 44.4%
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The ScreenASR-Bench paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| MiMo-Audio-7B-Instruct | Audio only | 9.63% | |||
| MiMo-V2.5 | Audio only | 9.25% | |||
| MiMo-V2.5 | Audio + Image | 7.11% | |||
| Audio only | 6.87% | ||||
| MiMo-Audio-7B-Instruct | Audio + Caption | 6.52% | |||
| Audio + Caption | 5.81% | ||||
| Audio only | 5.79% | ||||
| Audio only | 5.12% | ||||
| MiMo-V2.5 | Audio + Caption | 4.83% | |||
| Audio + Image | 4.69% | ||||
| Audio + Image | 4.40% | ||||
| Audio + Caption | 3.94% | ||||
| Audio + Caption | 3.47% |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The ScreenASR-Bench paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| MiMo-Audio-7B-Instruct | Audio only | 9.63% | |||
| MiMo-V2.5 | Audio only | 9.25% | |||
| MiMo-V2.5 | Audio + Image | 7.11% | |||
| Audio only | 6.87% | ||||
| MiMo-Audio-7B-Instruct | Audio + Caption | 6.52% | |||
| Audio + Caption | 5.81% | ||||
| Audio only | 5.79% | ||||
| Audio only | 5.12% | ||||
| MiMo-V2.5 | Audio + Caption | 4.83% | |||
| Audio + Image | 4.69% | ||||
| Audio + Image | 4.40% | ||||
| Audio + Caption | 3.94% | ||||
| Audio + Caption | 3.47% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The ScreenASR-Bench paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official ScreenASR-Bench leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 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%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 7 months ago, in April 2026, for Doubao Seed 2.0 Lite. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (12 entries).
- Pass Models scored 5 models we track, plus 7 other entries on its leaderboard.
- Warning Newest score The newest came 7 months ago, in April 2026, for Doubao Seed 2.0 Lite.
- Note Frontier models None of its models are on the capabilities index.
- Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 13 rows for models we track and 7 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.