VELOCITI
Models distinguish a video's correct description from a caption that assigns an action to the wrong person.
The test pairs captions with video clips to check who did what.
- Made by
- Darshana Saravanan and colleagues
- Released
- June 2024
- Kind
- Video understanding
- Level
- School
- Scored
- Checked automatically
- Answers
- Prediction
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision
- Last model added
- August 2024
- Scores refreshed
- November 20th, 2026
LLaVA-OneVision 72B leads with 43.1% among models released by August 2024 LLaVA-OneVision 72B leads with 43.1% among models released by August 2024
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. This view shows only models we track, each at its best.
The newest model with a score here came out in August 2024, 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 LLaVA-OneVision 72B 43.1%
- 2
LLaVA-OneVision 7B 32.1%
- 3 Qwen2-VL-7B-Instruct 15.9%
- 4 Video-LLaVA-7B 7.30%
- 5 OwlCon 1.70%
- 6 P-LLaVA 0%
Source: Official VELOCITI leaderboard
- 1 LLaVA-OneVision 72B 43.1%
- 2
LLaVA-OneVision 7B 32.1%
- 3 Qwen2-VL-7B-Instruct 15.9%
- 4 Video-LLaVA-7B 7.30%
Source: Official VELOCITI leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | LLaVA-OneVision 72B | 43.1% | Aug 2024 | Official leaderboard | |
| 2 | | 32.1% | Aug 2024 | Official leaderboard | |
| 3 | Qwen2-VL-7B-Instruct | 15.9% | Aug 2024 | Official leaderboard | |
| 4 | Video-LLaVA-7B | 7.30% | Nov 2023 | Official leaderboard | |
| 5 | OwlCon | 1.70% | Official leaderboard | ||
| 6 | P-LLaVA | 0% | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | LLaVA-OneVision 72B | 43.1% | Aug 2024 | Official leaderboard | |
| 2 | | 32.1% | Aug 2024 | Official leaderboard | |
| 3 | Qwen2-VL-7B-Instruct | 15.9% | Aug 2024 | Official leaderboard | |
| 4 | Video-LLaVA-7B | 7.30% | Nov 2023 | Official leaderboard |
Rankings by part
We read 8 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Control, Agent Random and Agent Binding. LLaVA-OneVision 72B leads 6 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.
Control
4 models we track have a score here.
- 1 Qwen2-VL-7B-Instruct ↑2 84.6%
- 2
LLaVA-OneVision 7B 81.6%
- 3 LLaVA-OneVision 72B ↓2 79.3%
- 4 Video-LLaVA-7B 65.8%
Agent Random
4 models we track have a score here.
- 1 LLaVA-OneVision 72B 63.7%
- 2
LLaVA-OneVision 7B 56.7%
- 3 Qwen2-VL-7B-Instruct 39.1%
- 4 Video-LLaVA-7B 16.4%
Agent Binding
4 models we track have a score here.
- 1 LLaVA-OneVision 72B 45.4%
- 2
LLaVA-OneVision 7B 32.9%
- 3 Qwen2-VL-7B-Instruct 13.5%
- 4 Video-LLaVA-7B 7.60%
Agent Coreference
4 models we track have a score here.
- 1 LLaVA-OneVision 72B 38.6%
- 2
LLaVA-OneVision 7B 8.00%
- 3 Qwen2-VL-7B-Instruct 6.50%
- 4 Video-LLaVA-7B 0.30%
Action Adversarial
4 models we track have a score here.
- 1 LLaVA-OneVision 72B 33.1%
- 2
LLaVA-OneVision 7B 29.7%
- 3 Qwen2-VL-7B-Instruct 17.8%
- 4 Video-LLaVA-7B 8.70%
Action Mannerism
4 models we track have a score here.
- 1
LLaVA-OneVision 7B ↑1 30.6%
- 2 LLaVA-OneVision 72B ↓1 29.3%
- 3 Qwen2-VL-7B-Instruct 17.5%
- 4 Video-LLaVA-7B 3.30%
Action Binding
4 models we track have a score here.
- 1 LLaVA-OneVision 72B 45.1%
- 2
LLaVA-OneVision 7B 36.4%
- 3 Qwen2-VL-7B-Instruct 16.4%
- 4 Video-LLaVA-7B 10.6%
Event Chronology
4 models we track have a score here.
- 1 LLaVA-OneVision 72B 46.5%
- 2
LLaVA-OneVision 7B 30.5%
- 3 Video-LLaVA-7B ↑1 3.90%
- 4 Qwen2-VL-7B-Instruct ↓1 0.40%
2 models have held number one on the official leaderboard since its release in June 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official VELOCITI 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 VELOCITI 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
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in August 2024, for Qwen2-VL-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 3 pass 1 note
- Pass Independent runs Scores from the official leaderboard (6 entries).
- Pass Models scored 4 models we track, plus 2 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in August 2024, for Qwen2-VL-7B-Instruct.
- Note Frontier models None of its models are on the capabilities index.
- Pass Headroom The best is 43.1%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on katha-ai.github.io: 4 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.