Ego-MC-Bench
Models watch a cook follow spoken instructions and tell them when to correct a mistake. The evaluator checks intervention timing and recipe step completion.
The authors report scores for 12 models on individual recipe steps. Gemini 3 Flash scored 0.23 for mistake detection on steps and 0.09 on full recipes.
- Released
- June 2026
- Kind
- Video understanding
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Multi-turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision, Audio and speech
- Strongest model tested
-
Gemini 3 Flash #60 on the capability index
- Last model added
- February 2026
- Scores refreshed
- November 20th, 2026
Qwen3.5-2B (Ego-CoMist+) tops the leaderboard with 0.27 2 models share the top score of 0.23 among models released by February 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 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 2.5 Flash, at 0.23. This view shows only models we track, each at its best.
The newest model with a score here came out in February 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
- 0.27 Qwen3.5-2B (Ego-CoMist+)
- 0.23
Gemini 2.5 Flash
- 0.23
Gemini 3 Flash
- 0.20 ProAssist
- 0.20 Qwen3-VL-2B (Ego-CoMist+)
- 0.19 Qwen3.5-2B (Ego-CoMist)
- 0.18 Qwen3-VL-32B
- 0.17
Qwen3.5-27B
- 0.12
Qwen3.5-9B
- 0.11 Qwen3.5-2B (QICD)
- 0.08 Qwen3.5-0.8B (Ego-CoMist+)
- 0.06 LiveCC
- 0.05 Videollm-online
- 0.05
Qwen3.5-2B
- 0.00 Qwen3-VL-8B
- 0.00 VideoLLAMA3-7B
- 0.00
Qwen2.5-VL-32B
- 0.00
InternVL3.5-38B
- 1 Qwen3.5-2B (Ego-CoMist+) Per-recipe step 0.27
- 2
Gemini 2.5 Flash Sep 2025, Per-recipe step 0.23
- 3
Gemini 3 Flash Per-recipe step 0.23
- 4 ProAssist Per-recipe step 0.20
- 5 Qwen3-VL-2B (Ego-CoMist+) Per-recipe step 0.20
- 6 Qwen3.5-2B (Ego-CoMist) Per-recipe step 0.19
- 7 Qwen3-VL-32B Per-recipe step 0.18
- 8
Qwen3.5-27B Per-recipe step 0.17
- 9
Qwen3.5-9B Per-recipe step 0.12
- 10 Qwen3.5-2B (QICD) Per-recipe step 0.11
- 11 Qwen3.5-0.8B (Ego-CoMist+) Per-recipe step 0.08
- 12 LiveCC Per-recipe step 0.06
- 13 Videollm-online Per-recipe step 0.05
- 14
Qwen3.5-2B Per-recipe step 0.05
- 15 Qwen3-VL-8B Per-recipe step 0.00
- 16 VideoLLAMA3-7B Per-recipe step 0.00
- 17
Qwen2.5-VL-32B Per-recipe step 0.00
- 18
InternVL3.5-38B Per-recipe step 0.00
Source: Official Ego-MC-Bench leaderboard
- 1
Gemini 2.5 Flash Sep 2025, Per-recipe step 0.23
- 2
Gemini 3 Flash Per-recipe step 0.23
- 3
Qwen3.5-27B Per-recipe step 0.17
- 4
Qwen3.5-9B Per-recipe step 0.12
- 5
Qwen3.5-2B Per-recipe step 0.05
- 6 VideoLLAMA3-7B Per-recipe step 0.00
- 7
Qwen2.5-VL-32B Per-recipe step 0.00
- 8
InternVL3.5-38B Per-recipe step 0.00
Source: Official Ego-MC-Bench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Qwen3.5-2B (Ego-CoMist+) | 0.27 | Per-recipe step | Official leaderboard | |
| 2 | | 0.23 | Per-recipe step | Sep 2025 | Official leaderboard |
| 3 | | 0.23 | Per-recipe step | Dec 2025 | Official leaderboard |
| 4 | ProAssist | 0.20 | Per-recipe step | Official leaderboard | |
| 5 | Qwen3-VL-2B (Ego-CoMist+) | 0.20 | Per-recipe step | Official leaderboard | |
| 6 | Qwen3.5-2B (Ego-CoMist) | 0.19 | Per-recipe step | Official leaderboard | |
| 7 | Qwen3-VL-32B | 0.18 | Per-recipe step | Official leaderboard | |
| 8 | | 0.17 | Per-recipe step | Feb 2026 | Official leaderboard |
| 9 | | 0.12 | Per-recipe step | Feb 2026 | Official leaderboard |
| 10 | Qwen3.5-2B (QICD) | 0.11 | Per-recipe step | Official leaderboard | |
| 11 | Qwen3.5-0.8B (Ego-CoMist+) | 0.08 | Per-recipe step | Official leaderboard | |
| 12 | LiveCC | 0.06 | Per-recipe step | Official leaderboard | |
| 13 | Videollm-online | 0.05 | Per-recipe step | Official leaderboard | |
| 14 | | 0.05 | Per-recipe step | Feb 2026 | Official leaderboard |
| 15 | Qwen3-VL-8B | 0.00 | Per-recipe step | Official leaderboard | |
| 16 | VideoLLAMA3-7B | 0.00 | Per-recipe step | Jan 2024 | Official leaderboard |
| 17 | | 0.00 | Per-recipe step | Mar 2025 | Official leaderboard |
| 18 | | 0.00 | Per-recipe step | Aug 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 0.23 | Per-recipe step | Sep 2025 | Official leaderboard |
| 2 | | 0.23 | Per-recipe step | Dec 2025 | Official leaderboard |
| 3 | | 0.17 | Per-recipe step | Feb 2026 | Official leaderboard |
| 4 | | 0.12 | Per-recipe step | Feb 2026 | Official leaderboard |
| 5 | | 0.05 | Per-recipe step | Feb 2026 | Official leaderboard |
| 6 | VideoLLAMA3-7B | 0.00 | Per-recipe step | Jan 2024 | Official leaderboard |
| 7 | | 0.00 | Per-recipe step | Mar 2025 | Official leaderboard |
| 8 | | 0.00 | Per-recipe step | Aug 2025 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Ego-MC-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 Ego-MC-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 9 months ago, in February 2026, for Qwen3.5-27B. 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 official leaderboard (18 entries).
- Pass Models scored 8 models we track, plus 10 other entries on its leaderboard.
- Warning Newest score The newest came 9 months ago, in February 2026, for Qwen3.5-27B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3 Flash (#60).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on apratimbh.github.io: 10 rows for models we track and 10 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.