Benchmark Almanac

ContextASR-Bench

Kimi-Audio-7B-Instruct 3.10% leads since July 2025
Not in model cards

Models transcribe speech with optional domain labels and named entities. The benchmark measures transcription errors and recognition of named entities.

Made by
He Wang and colleagues
Released
July 2025
Kind
Speech recognition and translation
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
In and out
Audio, Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Audio and speech
Last model added
April 2025
Scores refreshed
November 20th, 2026
12
Entries in its paper, 7 models we track

Kimi-Audio-7B-Instruct leads with 3.10% among models released by April 2025 Kimi-Audio-7B-Instruct leads with 3.10% among models released by April 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The headline averages word error rate across the four language and speech-format combinations. 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 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Kimi-Audio-7B-Instruct Fine-grained context 3.10%
  2. 2 Qwen2.5-Omni-7B Fine-grained context 3.38%
  3. 3 Qwen2.5-Omni-3B Fine-grained context 3.52%
  4. 4 Baichuan-Audio-Instruct Fine-grained context 4.58%
  5. 5 FireredASR-LLM-L No context 4.88%
  6. 6 Baichuan-Omni-1.5 Fine-grained context 6.51%
  7. 7 FireredASR-AED-L No context 9.36%
  8. 8 Sensevoice-Small No context 9.88%
  9. 9 Whisper large-v3 No context 10.4%
  10. 10 Qwen2-Audio-7B-Instruct Fine-grained context 10.6%
  11. 11 Whisper large-v3-turbo No context 11%
  12. 12 Paraformer-Large No context 18.4%

Source: The ContextASR-Bench paper

#EntryScoreSettingDateSource
1 Kimi-Audio-7B-Instruct 3.10% Fine-grained context Apr 2025 Official leaderboard
2 Qwen2.5-Omni-7B 3.38% Fine-grained context Mar 2025 Official leaderboard
3 Qwen2.5-Omni-3B 5.4B 3.52% Fine-grained context Official leaderboard
4 Baichuan-Audio-Instruct 4.58% Fine-grained context Feb 2025 Official leaderboard
5 FireredASR-LLM-L 8.3B 4.88% No context Official leaderboard
6 Baichuan-Omni-1.5 6.51% Fine-grained context Jan 2025 Official leaderboard
7 FireredASR-AED-L 1.1B 9.36% No context Official leaderboard
8 Sensevoice-Small 0.2B 9.88% No context Official leaderboard
9 Whisper large-v3 10.4% No context Nov 2023 Official leaderboard
10 Qwen2-Audio-7B-Instruct 10.6% Fine-grained context Aug 2024 Official leaderboard
11 Whisper large-v3-turbo 11% No context Oct 2024 Official leaderboard
12 Paraformer-Large 0.2B 18.4% No context Official leaderboard
#ModelScoreSettingDateSource
1 Kimi-Audio-7B-Instruct 3.10% Fine-grained context Apr 2025 Official leaderboard
2 Qwen2.5-Omni-7B 3.38% Fine-grained context Mar 2025 Official leaderboard
3 Baichuan-Audio-Instruct 4.58% Fine-grained context Feb 2025 Official leaderboard
4 Baichuan-Omni-1.5 6.51% Fine-grained context Jan 2025 Official leaderboard
5 Whisper large-v3 10.4% No context Nov 2023 Official leaderboard
6 Qwen2-Audio-7B-Instruct 10.6% Fine-grained context Aug 2024 Official leaderboard
7 Whisper large-v3-turbo 11% No context Oct 2024 Official leaderboard

Rankings by part

We read 12 parts from the benchmark’s paper, each ranking models on one part of the test, such as Speech-EN: WER, Speech-EN: NE-WER and Speech-EN: NE-FNR. Baichuan-Audio-Instruct leads 8 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Speech-EN: WER

7 models we track have a score here.

  1. 1 Kimi-Audio-7B-Instruct 2.90%
  2. 2 Qwen2.5-Omni-7B 3.96%
  3. 3 Baichuan-Audio-Instruct 7.52%
  4. 4 Baichuan-Omni-1.5 8.16%
  5. 5 Whisper large-v3 9.36%
  6. 6 Whisper large-v3-turbo ↑1 9.84%
  7. 7 Qwen2-Audio-7B-Instruct ↓1 11.5%

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.

Lowest effort to highest, light to dark
Qwen2-Audio-7B-Instruct
Fine-graine… No context
11.3%
Baichuan-Omni-1.5
Fine-graine… Coarse-grai…
7.63%
Baichuan-Audio-Instruct
Fine-graine… Coarse-grai… No context
8.90%
Qwen2.5-Omni-7B
Fine-graine… No context
4.55%
Kimi-Audio-7B-Instruct
Fine-graine…
3.77%

Source: The ContextASR-Bench paper

ModelEffortHarnessScoreCost per taskOutput tokens
Qwen2-Audio-7B-Instruct No context 11.3%
Qwen2-Audio-7B-Instruct Coarse-grained context 11.3%
Whisper large-v3-turbo No context 11%
Qwen2-Audio-7B-Instruct Fine-grained context 10.6%
Whisper large-v3 No context 10.4%
Baichuan-Audio-Instruct No context 8.90%
Baichuan-Omni-1.5 No context 7.63%
Baichuan-Omni-1.5 Coarse-grained context 7.47%
Baichuan-Omni-1.5 Fine-grained context 6.51%
Baichuan-Audio-Instruct Coarse-grained context 5.67%
Baichuan-Audio-Instruct Fine-grained context 4.58%
Qwen2.5-Omni-7B Coarse-grained context 4.55%
Qwen2.5-Omni-7B No context 4.42%
Kimi-Audio-7B-Instruct Coarse-grained context 3.77%
Kimi-Audio-7B-Instruct No context 3.68%
Qwen2.5-Omni-7B Fine-grained context 3.38%
Kimi-Audio-7B-Instruct Fine-grained context 3.10%

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.

Lowest effort to highest, light to dark
Qwen2-Audio-7B-Instruct
Fine-graine… No context
11.3%
Baichuan-Omni-1.5
Fine-graine… Coarse-grai…
7.63%
Baichuan-Audio-Instruct
Fine-graine… Coarse-grai… No context
8.90%
Qwen2.5-Omni-7B
Fine-graine… No context
4.55%
Kimi-Audio-7B-Instruct
Fine-graine…
3.77%

Source: The ContextASR-Bench paper

ModelEffortHarnessScoreCost per taskOutput tokens
Qwen2-Audio-7B-Instruct No context 11.3%
Qwen2-Audio-7B-Instruct Coarse-grained context 11.3%
Whisper large-v3-turbo No context 11%
Qwen2-Audio-7B-Instruct Fine-grained context 10.6%
Whisper large-v3 No context 10.4%
Baichuan-Audio-Instruct No context 8.90%
Baichuan-Omni-1.5 No context 7.63%
Baichuan-Omni-1.5 Coarse-grained context 7.47%
Baichuan-Omni-1.5 Fine-grained context 6.51%
Baichuan-Audio-Instruct Coarse-grained context 5.67%
Baichuan-Audio-Instruct Fine-grained context 4.58%
Qwen2.5-Omni-7B Coarse-grained context 4.55%
Qwen2.5-Omni-7B No context 4.42%
Kimi-Audio-7B-Instruct Coarse-grained context 3.77%
Kimi-Audio-7B-Instruct No context 3.68%
Qwen2.5-Omni-7B Fine-grained context 3.38%
Kimi-Audio-7B-Instruct Fine-grained context 3.10%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Moonshot AI
Benchmark paper Not tested since Jul 2025

Source: The ContextASR-Bench paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2024 2025 Kimi-Audio-7B-Instruct Baichuan-Omni-1.5 Whisper large-v3 0% 20% 40% 60% 80% 100% 2024 2025 Kimi-Audio-7B-Instruct Whisper large-v3

Source: Official ContextASR-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

All 695 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 19 months ago, in April 2025, for Kimi-Audio-7B-Instruct. 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 7 models we track, plus 5 other entries on its leaderboard.
  • Warning Newest score The newest came 19 months ago, in April 2025, for Kimi-Audio-7B-Instruct.
  • 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: 17 rows for models we track and 5 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.

How we check