MotoVerse
Models use a rider's camera to anticipate maneuvers and predict nearby road users' trajectories. The authors compare vision, language, and conventional prediction models.
- Made by
- Varun A. Paturkar and colleagues
- Released
- November 2026
- Kind
- Embodied reasoning
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Prediction
- Tasks
- All public
- Measures
- Capability
- Language
- English
- In and out
- Video → Time series, Text
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision
- Last model added
- October 2025
- Scores refreshed
- November 20th, 2026
SlowFast tops the leaderboard with 51.1% Qwen3-VL-2B-Instruct leads with 46.3% among models released by October 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 is maneuver anticipation accuracy after observing the first 20% of the maneuver. 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 Qwen3-VL-2B-Instruct, at 46.3%. This view shows only models we track, each at its best.
The newest model with a score here came out in October 2025, 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 SlowFast Fine-tuned 51.1%
- 2 UniDriveVLA LoRA 50.4%
- 3 X3D Fine-tuned 49.3%
- 4 I3D Fine-tuned 48.5%
- 5 C2D Fine-tuned 48.1%
- 6
Qwen3-VL-2B-Instruct LoRA 46.3%
- 7
InternVL2.5-1B Zero-shot 31.5%
- 8
InternVL2.5-4B Zero-shot 30%
- 9 Cosmos-Reason2-8B Zero-shot 21.9%
- 10 DriveLMM-o1 Zero-shot 21.5%
- 11
Qwen3-VL-4B-Instruct Zero-shot 17%
- 12 Alpamayo-1.5-10B Zero-shot 9.00%
Source: The MotoVerse paper
- 1
Qwen3-VL-2B-Instruct LoRA 46.3%
- 2
InternVL2.5-1B Zero-shot 31.5%
- 3
InternVL2.5-4B Zero-shot 30%
- 4
Qwen3-VL-4B-Instruct Zero-shot 17%
Source: The MotoVerse paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | SlowFast | 51.1% | Fine-tuned | Official leaderboard | |
| 2 | UniDriveVLA | 50.4% | LoRA | Official leaderboard | |
| 3 | X3D | 49.3% | Fine-tuned | Official leaderboard | |
| 4 | I3D | 48.5% | Fine-tuned | Official leaderboard | |
| 5 | C2D | 48.1% | Fine-tuned | Official leaderboard | |
| 6 | | 46.3% | LoRA | Oct 2025 | Official leaderboard |
| 7 | | 31.5% | Zero-shot | Dec 2024 | Official leaderboard |
| 8 | | 30% | Zero-shot | Dec 2024 | Official leaderboard |
| 9 | Cosmos-Reason2-8B | 21.9% | Zero-shot | Official leaderboard | |
| 10 | DriveLMM-o1 | 21.5% | Zero-shot | Official leaderboard | |
| 11 | | 17% | Zero-shot | Oct 2025 | Official leaderboard |
| 12 | Alpamayo-1.5-10B | 9.00% | Zero-shot | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 46.3% | LoRA | Oct 2025 | Official leaderboard |
| 2 | | 31.5% | Zero-shot | Dec 2024 | Official leaderboard |
| 3 | | 30% | Zero-shot | Dec 2024 | Official leaderboard |
| 4 | | 17% | Zero-shot | Oct 2025 | Official leaderboard |
Rankings by part
We read 18 parts from the benchmark’s paper in 5 groups: 20% observation, 30% observation, 50% observation, 100% observation and trajectory prediction. Each ranks models on one part of the test. Qwen3-VL-2B-Instruct leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
Accuracy
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 46.3%
- 2
InternVL2.5-1B 31.5%
- 3
InternVL2.5-4B 30%
- 4
Qwen3-VL-4B-Instruct 17%
Weighted F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 44.5%
- 2
InternVL2.5-4B ↑1 18.7%
- 3
InternVL2.5-1B ↓1 17.9%
- 4
Qwen3-VL-4B-Instruct 13.2%
Macro F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 36%
- 2
InternVL2.5-4B ↑1 12.8%
- 3
Qwen3-VL-4B-Instruct ↑1 10.1%
- 4
InternVL2.5-1B ↓2 9.20%
Accuracy
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 47%
- 2
InternVL2.5-1B 32.2%
- 3
InternVL2.5-4B 31.5%
- 4
Qwen3-VL-4B-Instruct 22.6%
Weighted F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 45.5%
- 2
Qwen3-VL-4B-Instruct ↑2 20.3%
- 3
InternVL2.5-1B ↓1 18.5%
- 4
InternVL2.5-4B ↓1 17.8%
Macro F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 36.4%
- 2
Qwen3-VL-4B-Instruct ↑2 15.7%
- 3
InternVL2.5-4B 12.1%
- 4
InternVL2.5-1B ↓2 9.70%
Accuracy
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 67.8%
- 2
Qwen3-VL-4B-Instruct ↑2 34.1%
- 3
InternVL2.5-1B ↓1 32.2%
- 4
InternVL2.5-4B ↓1 32.2%
Weighted F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 64.9%
- 2
Qwen3-VL-4B-Instruct ↑2 34.1%
- 3
InternVL2.5-1B ↓1 18.5%
- 4
InternVL2.5-4B ↓1 17.2%
Macro F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 46.6%
- 2
Qwen3-VL-4B-Instruct ↑2 23.2%
- 3
InternVL2.5-4B 11.8%
- 4
InternVL2.5-1B ↓2 9.70%
Accuracy
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 80.7%
- 2
Qwen3-VL-4B-Instruct ↑2 39.3%
- 3
InternVL2.5-4B 31.5%
- 4
InternVL2.5-1B ↓2 30%
Weighted F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 78.2%
- 2
Qwen3-VL-4B-Instruct ↑2 39.9%
- 3
InternVL2.5-4B 17.7%
- 4
InternVL2.5-1B ↓2 15.6%
Macro F1
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 56.5%
- 2
Qwen3-VL-4B-Instruct ↑2 26%
- 3
InternVL2.5-4B 14.1%
- 4
InternVL2.5-1B ↓2 7.60%
ADE
Displacement in pixels. 4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 58.7
- 2
Qwen3-VL-4B-Instruct ↑2 70.0
- 3
InternVL2.5-4B 85.6
- 4
InternVL2.5-1B ↓2 103.4
FDE
Displacement in pixels. 4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 140.8
- 2
Qwen3-VL-4B-Instruct ↑2 159.1
- 3
InternVL2.5-4B 179.1
- 4
InternVL2.5-1B ↓2 194.2
sADE
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 0.44
- 2
Qwen3-VL-4B-Instruct ↑2 0.52
- 3
InternVL2.5-4B 0.62
- 4
InternVL2.5-1B ↓2 0.78
sFDE
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 0.94
- 2
Qwen3-VL-4B-Instruct ↑2 1.08
- 3
InternVL2.5-4B 1.17
- 4
InternVL2.5-1B ↓2 1.27
AIoU
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 0.52
- 2
Qwen3-VL-4B-Instruct ↑2 0.43
- 3
InternVL2.5-4B 0.37
- 4
InternVL2.5-1B ↓2 0.29
FIoU
4 models we track have a score here.
- 1
Qwen3-VL-2B-Instruct 0.24
- 2
Qwen3-VL-4B-Instruct ↑2 0.17
- 3
InternVL2.5-4B 0.15
- 4
InternVL2.5-1B ↓2 0.14
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The MotoVerse 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 MotoVerse 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 Robotics
- ERQA In use 400 multiple-choice visual questions from robot and egocentric footage on spatial, trajectory, action and state reasoning. Accuracy.
Gemini Robotics ER 2 78.5%
- SimplerEnv In use Real-to-sim replicas of Google Robot and WidowX/Bridge setups. Policies scored by success rate on pick, move, drawer and Bridge tasks. Praxis-1 80%
- RoboLab In use Zero-shot DROID Franka policies in Isaac Lab on 120 tabletop tasks at three instruction levels, 10 episodes each. Success rate. Praxis-1 46.2%
- RoboDojo In use Generalist manipulation policies run 42 Isaac Sim tasks and 18 remote real-robot tasks. Scored by task score and success rate. GPT-6 Astra 27.6%
- Progress classification (Google DeepMind) In use Models assign each frame of a robot video to one of five task-progress bands from 0 to 100 percent. Scored by accuracy.
Gemini Robotics ER 2 57.4%
- Human proximity (Google DeepMind robotics) In use Models judge from robot camera views whether a human is within 1 meter, a spatial awareness check for safe robot behavior. Scored by accuracy.
Gemini Robotics ER 2 93%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 13 months ago, in October 2025, for Qwen3-VL-2B-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 benchmark’s paper (12 entries).
- Pass Models scored 4 models we track, plus 8 other entries on its leaderboard.
- Warning Newest score The newest came 13 months ago, in October 2025, for Qwen3-VL-2B-Instruct.
- Note Frontier models None of its models are on the capabilities index.
- Pass Headroom The best is 46.3%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 5 rows for models we track and 8 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.