Benchmark
RealWorldQA
RealWorldQA is a benchmark designed to evaluate basic real-world spatial understanding capabilities of multimodal models. The initial release consists of…
- Modality
- multimodal
- Categories
- spatial_reasoning, vision
- Openness
- unknown
- Source
- llm_stats
- Reported scores
- 29
Reported scores
llm_stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| DeepSeek VL2 | DeepSeek | 0.684 | 2024-12-13 | evidence |
| DeepSeek VL2 Small | DeepSeek | 0.654 | 2024-12-13 | evidence |
| DeepSeek VL2 Tiny | DeepSeek | 0.642 | 2024-12-13 | evidence |
| Grok-1.5V | xAI | 0.687 | 2024-04-12 | evidence |
| LFM2.5-VL-3B | Liquid AI | 0.731 | 2026-08-12 | evidence |
| North Micro Vision Instruct | Cohere | 0.622 | 2026-08-12 | evidence |
| Qwen2-VL-72B-Instruct | Qwen | 0.778 | 2024-08-29 | evidence |
| Qwen2.5-Omni-7B | Qwen | 0.703 | 2025-03-27 | evidence |
| Qwen3 VL 235B A22B Instruct | Qwen | 0.793 | 2025-09-22 | evidence |
| Qwen3 VL 235B A22B Thinking | Qwen | 0.813 | 2025-09-22 | evidence |
| Qwen3 VL 30B A3B Instruct | Qwen | 0.737 | 2025-09-22 | evidence |
| Qwen3 VL 30B A3B Thinking | Qwen | 0.774 | 2025-09-22 | evidence |
| Qwen3 VL 32B Instruct | Qwen | 0.79 | 2025-09-22 | evidence |
| Qwen3 VL 32B Thinking | Qwen | 0.784 | 2025-09-22 | evidence |
| Qwen3 VL 4B Instruct | Qwen | 0.709 | 2025-09-22 | evidence |
| Qwen3 VL 4B Thinking | Qwen | 0.732 | 2025-09-22 | evidence |
| Qwen3 VL 8B Instruct | Qwen | 0.715 | 2025-09-22 | evidence |
| Qwen3 VL 8B Thinking | Qwen | 0.735 | 2025-09-22 | evidence |
| Qwen3.5-122B-A10B | Qwen | 0.851 | 2026-02-24 | evidence |
| Qwen3.5-27B | Qwen | 0.837 | 2026-02-24 | evidence |
| Qwen3.5-35B-A3B | Qwen | 0.841 | 2026-02-24 | evidence |
| Qwen3.6 Plus | Qwen | 0.854 | 2026-04-02 | evidence |
| Qwen3.6-27B | Qwen | 0.841 | 2026-04-21 | evidence |
| Qwen3.6-35B-A3B | Qwen | 0.853 | 2026-04-16 | evidence |
| Qwen3.7-Plus | Qwen | 0.869 | 2026-05-31 | evidence |
| Qwen3.8 Max | Qwen | 0.88 | 2026-08-02 | evidence |
| Qwen3.8-27B | Qwen | 0.859 | 2026-08-14 | evidence |
| Seed 2.1 Pro | ByteDance | 0.867 | 2026-06-24 | evidence |
| Seed 2.1 Turbo | ByteDance | 0.863 | 2026-06-24 | 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.