RSLiveBench-UAV
Models answer questions about vehicles in drone images. The benchmark checks perception and spatial reasoning across different views and adds newly captured scenes.
- Made by
- East China Normal University and collaborators
- Released
- November 2026
- Kind
- Embodied reasoning
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- New items on a schedule
- Run by
- A host
- In views
- Vision, Kept fresh
- Strongest model tested
- GPT-6 Astra #4 on the capability index
- Last model added
- October 2026
- Scores refreshed
- November 20th, 2026
GPT-6 Astra leads with 59.9 GPT-6 Astra leads with 59.9
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. 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.
- 59.9 GPT-6 Astra
- 59.1 GPT-6.1 Sol
- 43.3 ACE-Brain-0.6-9B
- 37.8 GPT-5.6 Luna
- 37.2
DeepSeek V4.1 Flash 1017
- 36.1
Qwen3.5-9B
- 35.1 RynnBrain1.1-9B
- 32.8 RoboBrain2.5-8B-NV
- 31.7 RynnBrain-8B
- 31.2
Qwen3-VL-8B-Instruct
- 24.1 PhysBrain1.5-8B
- 24.0 MiMo-Embodied-7B
- 21.1 ACE-Brain-0-8B
- 20.8 Cosmos-Reason2-8B
- 18.9 SenseNova-SI-1.4-InternVL3-8B
- 18.8
InternVL3.5-4B
- 17.6
InternVL3.5-8B
- 16.9 SenseNova-SI-1.5-InternVL3-8B
- 16.6
SenseNova-SI-1.3-InternVL3-8B
- 15.2
Qwen3.5-2B
- 14.9 ACE-Brain-0.5-8B
- 14.8 SenseNova-SI-1.1-InternVL3-8B
- 14.5 RynnBrain1.1-2B
- 14.3 Cosmos-Reason2-2B
- 14.1 RynnBrain-2B
- 11.9 Pelican1.0-VL-7B
- 9.13 SenseNova-SI-1.1-InternVL3-2B
- 8.31 GeoChat-7B
- 4.38 VHM
- 0.11 GeoLLaVA-8K
- 1 GPT-6 Astra Low reasoning effort, Preliminary 59.9
- 2 GPT-6.1 Sol Low reasoning effort, Preliminary 59.1
- 3 ACE-Brain-0.6-9B Preliminary 43.3
- 4 GPT-5.6 Luna Low reasoning effort, Preliminary 37.8
- 5
DeepSeek V4.1 Flash 1017 Low reasoning effort, Preliminary 37.2
- 6
Qwen3.5-9B Preliminary 36.1
- 7 RynnBrain1.1-9B Preliminary 35.1
- 8 RoboBrain2.5-8B-NV Preliminary 32.8
- 9 RynnBrain-8B Preliminary 31.7
- 10
Qwen3-VL-8B-Instruct Preliminary 31.2
- 11 PhysBrain1.5-8B Preliminary 24.1
- 12 MiMo-Embodied-7B Preliminary 24.0
- 13 ACE-Brain-0-8B Preliminary 21.1
- 14 Cosmos-Reason2-8B Preliminary 20.8
- 15 SenseNova-SI-1.4-InternVL3-8B Preliminary 18.9
- 16
InternVL3.5-4B Preliminary 18.8
- 17
InternVL3.5-8B Preliminary 17.6
- 18 SenseNova-SI-1.5-InternVL3-8B Preliminary 16.9
- 19
SenseNova-SI-1.3-InternVL3-8B Preliminary 16.6
- 20
Qwen3.5-2B Preliminary 15.2
- 21 ACE-Brain-0.5-8B Preliminary 14.9
- 22 SenseNova-SI-1.1-InternVL3-8B Preliminary 14.8
- 23 RynnBrain1.1-2B Preliminary 14.5
- 24 Cosmos-Reason2-2B Preliminary 14.3
- 25 RynnBrain-2B Preliminary 14.1
- 26 Pelican1.0-VL-7B 2K, Preliminary 11.9
- 27 SenseNova-SI-1.1-InternVL3-2B Preliminary 9.13
- 28 GeoChat-7B Preliminary 8.31
- 29 VHM Preliminary 4.38
- 30 GeoLLaVA-8K Preliminary 0.11
Source: The RSLiveBench-UAV paper
- 1 GPT-6 Astra Low reasoning effort, Preliminary 59.9
- 2 GPT-6.1 Sol Low reasoning effort, Preliminary 59.1
- 3 GPT-5.6 Luna Low reasoning effort, Preliminary 37.8
- 4
DeepSeek V4.1 Flash 1017 Low reasoning effort, Preliminary 37.2
- 5
Qwen3.5-9B Preliminary 36.1
- 6
Qwen3-VL-8B-Instruct Preliminary 31.2
- 7
InternVL3.5-4B Preliminary 18.8
- 8
InternVL3.5-8B Preliminary 17.6
- 9
SenseNova-SI-1.3-InternVL3-8B Preliminary 16.6
- 10
Qwen3.5-2B Preliminary 15.2
Source: The RSLiveBench-UAV paper
| # | Entry | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra | 59.9 | Low reasoning effort, Preliminary | Sep 2026 | Official leaderboard | |
| 2 | GPT-6.1 Sol | 59.1 | Low reasoning effort, Preliminary | Sep 2026 | Official leaderboard | |
| 3 | ACE-Brain-0.6-9B open weights, 9B | 43.3 | Preliminary | huggingface.co | Official leaderboard | |
| 4 | GPT-5.6 Luna | 37.8 | Low reasoning effort, Preliminary | Jul 2026 | Official leaderboard | |
| 5 | | 37.2 | Low reasoning effort, Preliminary | Oct 2026 | deepseek.com | Official leaderboard |
| 6 | | 36.1 | Preliminary | Feb 2026 | huggingface.co | Official leaderboard |
| 7 | RynnBrain1.1-9B open weights, 9B | 35.1 | Preliminary | huggingface.co | Official leaderboard | |
| 8 | RoboBrain2.5-8B-NV open weights, 8B | 32.8 | Preliminary | huggingface.co | Official leaderboard | |
| 9 | RynnBrain-8B open weights, 8B | 31.7 | Preliminary | huggingface.co | Official leaderboard | |
| 10 | | 31.2 | Preliminary | Oct 2025 | huggingface.co | Official leaderboard |
| 11 | PhysBrain1.5-8B open weights, 8B | 24.1 | Preliminary | deepcybo-physai.github.io | Official leaderboard | |
| 12 | MiMo-Embodied-7B open weights, 7B | 24.0 | Preliminary | github.com | Official leaderboard | |
| 13 | ACE-Brain-0-8B open weights, 8B | 21.1 | Preliminary | huggingface.co | Official leaderboard | |
| 14 | Cosmos-Reason2-8B open weights, 8B | 20.8 | Preliminary | docs.nvidia.com | Official leaderboard | |
| 15 | SenseNova-SI-1.4-InternVL3-8B open weights, 8B | 18.9 | Preliminary | huggingface.co | Official leaderboard | |
| 16 | | 18.8 | Preliminary | Aug 2025 | huggingface.co | Official leaderboard |
| 17 | | 17.6 | Preliminary | Aug 2025 | huggingface.co | Official leaderboard |
| 18 | SenseNova-SI-1.5-InternVL3-8B open weights, 8B | 16.9 | Preliminary | huggingface.co | Official leaderboard | |
| 19 | | 16.6 | Preliminary | Jan 2026 | huggingface.co | Official leaderboard |
| 20 | | 15.2 | Preliminary | Feb 2026 | huggingface.co | Official leaderboard |
| 21 | ACE-Brain-0.5-8B open weights, 8B | 14.9 | Preliminary | huggingface.co | Official leaderboard | |
| 22 | SenseNova-SI-1.1-InternVL3-8B open weights, 8B | 14.8 | Preliminary | huggingface.co | Official leaderboard | |
| 23 | RynnBrain1.1-2B open weights, 2B | 14.5 | Preliminary | huggingface.co | Official leaderboard | |
| 24 | Cosmos-Reason2-2B open weights, 2B | 14.3 | Preliminary | docs.nvidia.com | Official leaderboard | |
| 25 | RynnBrain-2B open weights, 2B | 14.1 | Preliminary | huggingface.co | Official leaderboard | |
| 26 | Pelican1.0-VL-7B open weights, 7B | 11.9 | 2K, Preliminary | github.com | Official leaderboard | |
| 27 | SenseNova-SI-1.1-InternVL3-2B open weights, 2B | 9.13 | Preliminary | huggingface.co | Official leaderboard | |
| 28 | GeoChat-7B open weights, 7B | 8.31 | Preliminary | arxiv.org | Official leaderboard | |
| 29 | VHM open weights, 7B | 4.38 | Preliminary | github.com | Official leaderboard | |
| 30 | GeoLLaVA-8K open weights, 7B | 0.11 | Preliminary | github.com | Official leaderboard |
| # | Model | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra | 59.9 | Low reasoning effort, Preliminary | Sep 2026 | Official leaderboard | |
| 2 | GPT-6.1 Sol | 59.1 | Low reasoning effort, Preliminary | Sep 2026 | Official leaderboard | |
| 3 | GPT-5.6 Luna | 37.8 | Low reasoning effort, Preliminary | Jul 2026 | Official leaderboard | |
| 4 | | 37.2 | Low reasoning effort, Preliminary | Oct 2026 | deepseek.com | Official leaderboard |
| 5 | | 36.1 | Preliminary | Feb 2026 | huggingface.co | Official leaderboard |
| 6 | | 31.2 | Preliminary | Oct 2025 | huggingface.co | Official leaderboard |
| 7 | | 18.8 | Preliminary | Aug 2025 | huggingface.co | Official leaderboard |
| 8 | | 17.6 | Preliminary | Aug 2025 | huggingface.co | Official leaderboard |
| 9 | | 16.6 | Preliminary | Jan 2026 | huggingface.co | Official leaderboard |
| 10 | | 15.2 | Preliminary | Feb 2026 | huggingface.co | Official leaderboard |
Rankings by part
We read 21 parts from the benchmark’s paper in 5 groups: suite, 2d perception, 3d perception, cross-view perception and reasoning. Each ranks models on one part of the test. GPT-6 Astra leads 12 of them, and 6 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
2D perception
10 models we track have a score here.
- 1 GPT-6.1 Sol ↑1 54.3%
- 2 GPT-6 Astra ↓1 54.2%
- 3
Qwen3-VL-8B-Instruct ↑3 42.2%
- 4
Qwen3.5-9B ↑1 40.2%
- 5
DeepSeek V4.1 Flash 1017 ↓1 36.3%
- 6
Qwen3.5-2B ↑4 33.8%
- 7 GPT-5.6 Luna ↓4 33.6%
- 8
InternVL3.5-8B 24.2%
- 9
SenseNova-SI-1.3-InternVL3-8B 20.5%
- 10
InternVL3.5-4B ↓3 20%
3D perception
10 models we track have a score here.
- 1 GPT-6 Astra 77.5%
- 2 GPT-6.1 Sol 76.7%
- 3
Qwen3.5-9B ↑2 69.3%
- 4 GPT-5.6 Luna ↓1 61.3%
- 5
DeepSeek V4.1 Flash 1017 ↓1 60.7%
- 6
Qwen3-VL-8B-Instruct 48%
- 7
InternVL3.5-4B 27.7%
- 8
InternVL3.5-8B 20.7%
- 9
SenseNova-SI-1.3-InternVL3-8B 15.5%
- 10
Qwen3.5-2B 13.3%
Cross-view perception
10 models we track have a score here.
- 1 GPT-6 Astra 69.1%
- 2 GPT-6.1 Sol 67%
- 3
DeepSeek V4.1 Flash 1017 ↑1 32.4%
- 4 GPT-5.6 Luna ↓1 32%
- 5
Qwen3.5-9B 15.7%
- 6
Qwen3-VL-8B-Instruct 15.5%
- 7
InternVL3.5-8B ↑1 14.5%
- 8
SenseNova-SI-1.3-InternVL3-8B ↑1 13.8%
- 9
InternVL3.5-4B ↓2 11.6%
- 10
Qwen3.5-2B 6.05%
Reasoning
10 models we track have a score here.
- 1 GPT-6 Astra 38.8%
- 2 GPT-6.1 Sol 38.6%
- 3 GPT-5.6 Luna 24.3%
- 4
DeepSeek V4.1 Flash 1017 19.3%
- 5
Qwen3.5-9B 19.1%
- 6
Qwen3-VL-8B-Instruct 18.9%
- 7
SenseNova-SI-1.3-InternVL3-8B ↑2 16.4%
- 8
InternVL3.5-4B ↓1 16%
- 9
InternVL3.5-8B ↓1 10.8%
- 10
Qwen3.5-2B 7.48%
Brand/model recognition (per image)
10 models we track have a score here.
- 1 GPT-6 Astra 14%
- 2
Qwen3.5-9B ↑3 13.9%
- 3 GPT-6.1 Sol ↓1 13.5%
- 4
Qwen3.5-2B ↑6 12%
- 5
Qwen3-VL-8B-Instruct ↑1 11.8%
- 6
DeepSeek V4.1 Flash 1017 ↓2 11.1%
- 7 GPT-5.6 Luna ↓4 8.57%
- 8
InternVL3.5-8B 4.79%
- 9
InternVL3.5-4B ↓2 4.73%
- 10
SenseNova-SI-1.3-InternVL3-8B ↓1 4.21%
Brand/model recognition (per vehicle)
10 models we track have a score here.
- 1 GPT-6 Astra 26.3%
- 2 GPT-6.1 Sol 24.9%
- 3
Qwen3.5-9B ↑2 20.3%
- 4
Qwen3-VL-8B-Instruct ↑2 17.2%
- 5
DeepSeek V4.1 Flash 1017 ↓1 17.1%
- 6
Qwen3.5-2B ↑4 14.6%
- 7 GPT-5.6 Luna ↓4 8.00%
- 8
InternVL3.5-4B ↓1 4.92%
- 9
InternVL3.5-8B ↓1 4.77%
- 10
SenseNova-SI-1.3-InternVL3-8B ↓1 4.31%
Single-image vehicle counting
10 models we track have a score here.
- 1
InternVL3.5-8B ↑7 66.9%
- 2
Qwen3-VL-8B-Instruct ↑4 63.5%
- 3
DeepSeek V4.1 Flash 1017 ↑1 61.7%
- 4
Qwen3.5-9B ↑1 57.5%
- 5 GPT-6 Astra ↓4 57.2%
- 6
InternVL3.5-4B ↑1 56.6%
- 7 GPT-6.1 Sol ↓5 55.6%
- 8 GPT-5.6 Luna ↓5 50.6%
- 9
SenseNova-SI-1.3-InternVL3-8B 50%
- 10
Qwen3.5-2B 42.3%
2D spatial relations
10 models we track have a score here.
- 1 GPT-6.1 Sol ↑1 82.9%
- 2 GPT-6 Astra ↓1 82.5%
- 3
Qwen3.5-9B ↑2 74.7%
- 4
Qwen3-VL-8B-Instruct ↑2 73.7%
- 5
DeepSeek V4.1 Flash 1017 ↓1 67.1%
- 6
Qwen3.5-2B ↑4 66.3%
- 7 GPT-5.6 Luna ↓4 46.2%
- 8
SenseNova-SI-1.3-InternVL3-8B ↑1 27.2%
- 9
InternVL3.5-4B ↓2 0%
- 10
InternVL3.5-8B ↓2 0%
Weather understanding
10 models we track have a score here.
- 1
Qwen3-VL-8B-Instruct ↑5 76.9%
- 2 GPT-6.1 Sol 69.6%
- 3
Qwen3.5-9B ↑2 68.1%
- 4 GPT-6 Astra ↓3 67.7%
- 5
Qwen3.5-2B ↑5 60.8%
- 6
InternVL3.5-8B ↑2 55.6%
- 7 GPT-5.6 Luna ↓4 54.4%
- 8
DeepSeek V4.1 Flash 1017 ↓4 46%
- 9
InternVL3.5-4B ↓2 43.3%
- 10
SenseNova-SI-1.3-InternVL3-8B ↓1 34%
2D vehicle heading
10 models we track have a score here.
- 1 GPT-6.1 Sol ↑1 79.1%
- 2 GPT-6 Astra ↓1 77.3%
- 3 GPT-5.6 Luna 33.6%
- 4
DeepSeek V4.1 Flash 1017 15%
- 5
InternVL3.5-8B ↑3 12.9%
- 6
InternVL3.5-4B ↑1 10.2%
- 7
Qwen3-VL-8B-Instruct ↓1 10.1%
- 8
Qwen3.5-2B ↑2 6.70%
- 9
Qwen3.5-9B ↓4 6.70%
- 10
SenseNova-SI-1.3-InternVL3-8B ↓1 3.35%
Vehicle length and width
10 models we track have a score here.
- 1 GPT-6 Astra 97.4%
- 2 GPT-6.1 Sol 97.1%
- 3
Qwen3.5-9B ↑2 96.7%
- 4
DeepSeek V4.1 Flash 1017 96.1%
- 5 GPT-5.6 Luna ↓2 95.6%
- 6
InternVL3.5-4B ↑1 70%
- 7
InternVL3.5-8B ↑1 53.7%
- 8
Qwen3-VL-8B-Instruct ↓2 50.1%
- 9
SenseNova-SI-1.3-InternVL3-8B 5.12%
- 10
Qwen3.5-2B 0.48%
Vehicle height
10 models we track have a score here.
- 1 GPT-6 Astra 95.4%
- 2 GPT-6.1 Sol 94.8%
- 3
Qwen3.5-9B ↑2 91.8%
- 4 GPT-5.6 Luna ↓1 91.6%
- 5
DeepSeek V4.1 Flash 1017 ↓1 90.2%
- 6
Qwen3-VL-8B-Instruct 88.7%
- 7
SenseNova-SI-1.3-InternVL3-8B ↑2 42%
- 8
InternVL3.5-4B ↓1 40.9%
- 9
InternVL3.5-8B ↓1 20.9%
- 10
Qwen3.5-2B 4.22%
Relative spatial distance
10 models we track have a score here.
- 1
Qwen3.5-9B ↑4 51.8%
- 2
Qwen3-VL-8B-Instruct ↑4 48.7%
- 3 GPT-6.1 Sol ↓1 48.3%
- 4 GPT-6 Astra ↓3 46.5%
- 5
Qwen3.5-2B ↑5 42.1%
- 6
DeepSeek V4.1 Flash 1017 ↓2 35.1%
- 7 GPT-5.6 Luna ↓4 14.9%
- 8
SenseNova-SI-1.3-InternVL3-8B ↑1 14%
- 9
InternVL3.5-4B ↓2 0%
- 10
InternVL3.5-8B ↓2 0%
Absolute spatial distance
10 models we track have a score here.
- 1 GPT-6 Astra 70.6%
- 2 GPT-6.1 Sol 66.4%
- 3 GPT-5.6 Luna 42.8%
- 4
Qwen3.5-9B ↑1 37%
- 5
DeepSeek V4.1 Flash 1017 ↓1 21.4%
- 6
InternVL3.5-8B ↑2 8.41%
- 7
Qwen3.5-2B ↑3 6.48%
- 8
Qwen3-VL-8B-Instruct ↓2 4.46%
- 9
SenseNova-SI-1.3-InternVL3-8B 0.92%
- 10
InternVL3.5-4B ↓3 0.03%
Cross-view vehicle counting
10 models we track have a score here.
- 1 GPT-5.6 Luna ↑2 51.1%
- 2 GPT-6 Astra ↓1 49.3%
- 3 GPT-6.1 Sol ↓1 48.7%
- 4
InternVL3.5-8B ↑4 43.4%
- 5
DeepSeek V4.1 Flash 1017 ↓1 40.7%
- 6
InternVL3.5-4B ↑1 34.7%
- 7
SenseNova-SI-1.3-InternVL3-8B ↑2 34%
- 8
Qwen3-VL-8B-Instruct ↓2 33.4%
- 9
Qwen3.5-9B ↓4 27.1%
- 10
Qwen3.5-2B 10.1%
Cross-view grounding
10 models we track have a score here.
- 1 GPT-6 Astra 61.1%
- 2 GPT-6.1 Sol 57.2%
- 3
DeepSeek V4.1 Flash 1017 ↑1 14.2%
- 4 GPT-5.6 Luna ↓1 12.7%
- 5
Qwen3-VL-8B-Instruct ↑1 9.29%
- 6
SenseNova-SI-1.3-InternVL3-8B ↑3 5.20%
- 7
Qwen3.5-2B ↑3 4.30%
- 8
Qwen3.5-9B ↓3 2.01%
- 9
InternVL3.5-4B ↓2 0%
- 10
InternVL3.5-8B ↓2 0%
Instance re-identification
10 models we track have a score here.
- 1 GPT-6 Astra 96.8%
- 2 GPT-6.1 Sol 95%
- 3
DeepSeek V4.1 Flash 1017 ↑1 42.3%
- 4 GPT-5.6 Luna ↓1 32%
- 5
Qwen3.5-9B 17.9%
- 6
Qwen3-VL-8B-Instruct 3.95%
- 7
Qwen3.5-2B ↑3 3.74%
- 8
SenseNova-SI-1.3-InternVL3-8B ↑1 2.19%
- 9
InternVL3.5-4B ↓2 0%
- 10
InternVL3.5-8B ↓2 0%
Camera height and pitch
10 models we track have a score here.
- 1 GPT-6 Astra 66.7%
- 2 GPT-6.1 Sol 64.3%
- 3 GPT-5.6 Luna 50.1%
- 4
InternVL3.5-4B ↑3 30.1%
- 5
DeepSeek V4.1 Flash 1017 ↓1 28.8%
- 6
Qwen3.5-9B ↓1 28.5%
- 7
InternVL3.5-8B ↑1 26.9%
- 8
Qwen3-VL-8B-Instruct ↓2 26.8%
- 9
SenseNova-SI-1.3-InternVL3-8B 14.5%
- 10
Qwen3.5-2B 14.3%
Vehicle-center polar coordinates
10 models we track have a score here.
- 1 GPT-6 Astra 53.3%
- 2 GPT-6.1 Sol 52%
- 3
DeepSeek V4.1 Flash 1017 ↑1 22%
- 4
InternVL3.5-4B ↑3 21%
- 5 GPT-5.6 Luna ↓2 19.9%
- 6
Qwen3-VL-8B-Instruct 15.3%
- 7
Qwen3.5-9B ↓2 10.4%
- 8
SenseNova-SI-1.3-InternVL3-8B ↑1 9.59%
- 9
Qwen3.5-2B ↑1 6.00%
- 10
InternVL3.5-8B ↓2 4.14%
Camera location
10 models we track have a score here.
- 1
Qwen3.5-9B ↑4 25.7%
- 2 GPT-6 Astra ↓1 25.2%
- 3 GPT-6.1 Sol ↓1 25%
- 4
Qwen3-VL-8B-Instruct ↑2 22.9%
- 5 GPT-5.6 Luna ↓2 17.5%
- 6
DeepSeek V4.1 Flash 1017 ↓2 14.2%
- 7
SenseNova-SI-1.3-InternVL3-8B ↑2 11.7%
- 8
Qwen3.5-2B ↑2 4.37%
- 9
InternVL3.5-8B ↓1 1.39%
- 10
InternVL3.5-4B ↓3 0.97%
Capture hour
10 models we track have a score here.
- 1
SenseNova-SI-1.3-InternVL3-8B ↑8 29.8%
- 2 GPT-6.1 Sol 13.1%
- 3
DeepSeek V4.1 Flash 1017 ↑1 12.3%
- 4
InternVL3.5-4B ↑3 11.9%
- 5
Qwen3.5-9B 11.9%
- 6
InternVL3.5-8B ↑2 10.8%
- 7
Qwen3-VL-8B-Instruct ↓1 10.6%
- 8 GPT-6 Astra ↓7 10%
- 9 GPT-5.6 Luna ↓6 9.58%
- 10
Qwen3.5-2B 5.21%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The RSLiveBench-UAV 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 RSLiveBench-UAV 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
All 9 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (30 entries).
- Pass Models scored 10 models we track, plus 20 other entries on its leaderboard.
- Pass Newest score The newest is from October 2026, for DeepSeek V4.1 Flash 1017.
- Note Frontier models Only 2 of the top 10 have a score. The strongest tested is GPT-6 Astra (#4).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on rslivebench.space: 10 rows for models we track and 20 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.