RefSpatial-Bench
Qwen3.5 397B-A17B 73.6% leads since Feb 2026 Fading Real images where the model points to an object or free space defined by multi-step spatial relations. Scored by mask hit.
No independent leaderboard. Subset aggregation not stated. One cross-checked row (RoboBrain2.0-7B) differs from its origin: 42.2 vs 32.50.
- Level
- Everyday Most adults, without training
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Short answer A number, word or short string matched to a key.
Qwen3.5 397B-A17B leads with 73.6% among models released by Apr 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Apr 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 Qwen3.5 397B-A17B 73.6%
- 2
Seed2.0 Pro 72.6% - 3 Qwen3.6 27B 70%
- 4 Qwen3-VL-235B-A22B 69.9%
- 5 Qwen3.5-122B-A10B 69.3%
- 6 Qwen 3.6 35B-A3B 64.3%
- 7 Seed1.8 56.3%
- 8
gemini-robotics-er-1.5-preview 48.5%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Qwen3.5 397B-A17B | 73.6% | Feb 2026 | Model card | |
| 2 | Seed2.0 Pro | 72.6% | Feb 2026 | Model card | |
| 3 | Qwen3.6 27B | 70% | Apr 2026 | Model card | |
| 4 | Qwen3-VL-235B-A22B | 69.9% | Sep 2025 | Model card | |
| 5 | Qwen3.5-122B-A10B | 69.3% | Feb 2026 | Model card | |
| 6 | Qwen 3.6 35B-A3B | 64.3% | Apr 2026 | Model card | |
| 7 | Seed1.8 | 56.3% | Dec 2025 | Model card | |
| 8 | gemini-robotics-er-1.5-preview | 48.5% | Sep 2025 | Model card |
3 models have held number one since its release in Jun 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs headline it less often now, last in Apr 2026
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 22 Apr 2026 | Qwen3.6 27B | 70 | RefSpatialBench |
| 15 Apr 2026 | Qwen 3.6 35B-A3B | 64.3 | RefSpatialBench |
| 24 Feb 2026 | Qwen3.5-122B-A10B | 69.3 | RefSpatialBench |
| 16 Feb 2026 | Qwen3.5 397B-A17B | 73.6 | RefSpatialBench |
| 18 Dec 2025 | Seed1.8 | 56.3 | RefSpatialBench |
| 23 Sep 2025 | Qwen3-VL-235B-A22B | 69.9 | RefSpatialBench |
4 system cards report it, from 3 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 21 Mar 2026 | Seed1.8 model card | 56.3 | RefSpatialBench (Table 2) | Capabilities |
| 14 Feb 2026 | Seed2.0 Pro model card | 72.6 | RefSpatialBench (Table 8) | Capabilities |
| 26 Nov 2025 | Qwen3-VL-235B-A22B technical report | 69.9 | RefSpatialBench (Table 2) | Capabilities |
| 25 Sep 2025 | gemini-robotics-er-1.5-preview technical report | 48.5 | RefSpatial (Table 19, thinking on). Thinking off 41.8 | Capabilities |
Versions
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%
- RoboLab In use Zero-shot DROID Franka policies in Isaac Lab on 120 tabletop tasks at three instruction levels, 10 episodes each. Success rate.
GR00T N1.7 8.58% - 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%
- Moment finding (Google DeepMind) In use Models find the moment in a robot video when an event happens. Scored by accuracy and mean absolute time distance in seconds. Gemini Robotics ER 2 91.3%
- 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%
- LIBERO Fading Simulated Franka tabletop tasks in four 10-task suites. Policies fine-tuned on 50 demos per task. Success rate over ~50 trials.
MolmoAct 2 97.2%