Benchmark Almanac

MISHAP-Bench

Kimi-Audio-7B-Instruct 32.8% leads since September 2026
Not in model cards

Models answer open-ended questions about audio clips. A model judge uses human annotations to identify claims the recording does not support.

Made by
Zhi Wen Soi and colleagues
Released
September 2026
Kind
Audio understanding and voice
Level
Everyday
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Propensity
Behavior
Hallucination
Language
English
In and out
Text, Audio → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Audio and speech
Strongest model tested
Gemini 3.7 Flash #26 on the capability index
Last model added
August 2026
Scores refreshed
November 20th, 2026
10
Entries in its paper, 8 models we track

Kimi-Audio-7B-Instruct leads with 32.8% among models released by August 2026 Kimi-Audio-7B-Instruct leads with 32.8% among models released by August 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 August 2026, 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 AAD 32.8%
  2. 2 Qwen2.5-Omni-7B MTI 35.7%
  3. 3 Gemini 3.7 Flash W/O 36.5%
  4. 4 Covo-Audio AAD 43.6%
  5. 5 MiMo-V2.5 W/O 47.5%
  6. 6 Qwen2-Audio-7B-Instruct VCD 48.3%
  7. 7 Audio Flamingo 3 AAD 48.9%
  8. 8 Qwen3-Omni-30B-A3B AAD 54.3%
  9. 9 MiMo-Audio-7B-Instruct AAD 56%
  10. 10 Step-Audio 2 mini DoLa 66.7%

Source: The MISHAP-Bench paper

#EntryScoreSettingDateSource
1 Kimi-Audio-7B-Instruct 32.8% AAD Apr 2025 Official leaderboard
2 Qwen2.5-Omni-7B 35.7% MTI Mar 2025 Official leaderboard
3 Gemini 3.7 Flash 36.5% W/O Aug 2026 Official leaderboard
4 Covo-Audio 43.6% AAD Official leaderboard
5 MiMo-V2.5 47.5% W/O Apr 2026 Official leaderboard
6 Qwen2-Audio-7B-Instruct 48.3% VCD Aug 2024 Official leaderboard
7 Audio Flamingo 3 48.9% AAD Jul 2025 Official leaderboard
8 Qwen3-Omni-30B-A3B 54.3% AAD Sep 2025 Official leaderboard
9 MiMo-Audio-7B-Instruct 56% AAD Sep 2025 Official leaderboard
10 Step-Audio 2 mini 66.7% DoLa Official leaderboard
#ModelScoreSettingDateSource
1 Kimi-Audio-7B-Instruct 32.8% AAD Apr 2025 Official leaderboard
2 Qwen2.5-Omni-7B 35.7% MTI Mar 2025 Official leaderboard
3 Gemini 3.7 Flash 36.5% W/O Aug 2026 Official leaderboard
4 MiMo-V2.5 47.5% W/O Apr 2026 Official leaderboard
5 Qwen2-Audio-7B-Instruct 48.3% VCD Aug 2024 Official leaderboard
6 Audio Flamingo 3 48.9% AAD Jul 2025 Official leaderboard
7 Qwen3-Omni-30B-A3B 54.3% AAD Sep 2025 Official leaderboard
8 MiMo-Audio-7B-Instruct 56% AAD Sep 2025 Official leaderboard

Rankings by part

We read 18 parts from the benchmark’s paper in 5 groups: music, sound, speech, context and knowledge. Each ranks models on one part of the test. Kimi-Audio-7B-Instruct leads 7 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.

Context

8 models we track have a score here.

  1. 1 Qwen2.5-Omni-7B ↑1 50.3%
  2. 2 Kimi-Audio-7B-Instruct ↓1 50.6%
  3. 3 Audio Flamingo 3 ↑3 65.1%
  4. 4 Qwen2-Audio-7B-Instruct ↑1 67%
  5. 5 MiMo-Audio-7B-Instruct ↑3 67%
  6. 6 MiMo-V2.5 ↓2 72.1%
  7. 7 Gemini 3.7 Flash ↓4 73.6%
  8. 8 Qwen3-Omni-30B-A3B ↓1 74.6%

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
MiMo-Audio-7B-Instruct
AAD VCD
76.7%
Qwen3-Omni-30B-A3B
AAD DoLa
64.4%
Audio Flamingo 3
AAD VCD MTI
66.7%
Qwen2-Audio-7B-Instruct
VCD W/O
55.3%
Qwen2.5-Omni-7B
MTI
38.5%
Kimi-Audio-7B-Instruct
AAD DoLa W/O
51.3%

Source: The MISHAP-Bench paper

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
MiMo-Audio-7B-Instruct
AAD VCD
76.7%
Qwen3-Omni-30B-A3B
AAD DoLa
64.4%
Audio Flamingo 3
AAD VCD MTI
66.7%
Qwen2-Audio-7B-Instruct
VCD W/O
55.3%
Qwen2.5-Omni-7B
MTI
38.5%
Kimi-Audio-7B-Instruct
AAD DoLa W/O
51.3%

Source: The MISHAP-Bench paper

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 Sep 2026

Source: The MISHAP-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% 2025 2026 Kimi-Audio-7B-Instruct Qwen2.5-Omni-7B Qwen2-Audio-7B-Instruct Qwen2-Audio-7B-Instruct 0% 20% 40% 60% 80% 100% 2025 2026 Kimi-Audio-7B-Instruct Qwen2-Audio-7B-Instruct

Source: Official MISHAP-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

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (10 entries).
  • Pass Models scored 8 models we track, plus 2 other entries on its leaderboard.
  • Pass Newest score The newest is from August 2026, for Gemini 3.7 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3.7 Flash (#26).
  • 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: 32 rows for models we track and 2 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