Benchmark
RefSpatialBench
RefSpatialBench evaluates spatial reference understanding and grounding.
- Modality
- image
- Categories
- spatial_reasoning, grounding, vision
- Openness
- unknown
- Source
- llm_stats
- Reported scores
- 6
Reported scores
llm_stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Qwen3 VL 235B A22B Thinking | Qwen | 0.699 | 2025-09-22 | evidence |
| Qwen3.5-122B-A10B | Qwen | 0.693 | 2026-02-24 | evidence |
| Qwen3.5-27B | Qwen | 0.677 | 2026-02-24 | evidence |
| Qwen3.5-35B-A3B | Qwen | 0.635 | 2026-02-24 | evidence |
| Qwen3.6-27B | Qwen | 0.7 | 2026-04-21 | evidence |
| Qwen3.6-35B-A3B | Qwen | 0.643 | 2026-04-16 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.