spatial reasoning benchmark
RealWorldQA Leaderboard
RealWorldQA is a benchmark for evaluating basic real-world spatial understanding in multimodal models using over 700 anonymized images from vehicles and other real-world scenarios.
Updated Sep 5, 2026
Models31
Model coverage31
MetricScore
EvidenceB
RealWorldQA Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score88.50% | Percentile100.00% | Participants31 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score88.50% | Percentile96.67% | Participants31 | EvidenceC | Evaluated |
| Rank03 | ModelAC | Score88.00% | Percentile93.33% | Participants31 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score86.90% | Percentile90.00% | Participants31 | EvidenceC | Evaluated |
| Rank05 | ModelBY | Score86.70% | Percentile86.67% | Participants31 | EvidenceC | Evaluated |
| Rank06 | ModelBY | Score86.30% | Percentile83.33% | Participants31 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score85.90% | Percentile80.00% | Participants31 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score85.40% | Percentile76.67% | Participants31 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score85.30% | Percentile73.33% | Participants31 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score85.10% | Percentile70.00% | Participants31 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score84.10% | Percentile66.67% | Participants31 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score84.10% | Percentile63.33% | Participants31 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score83.70% | Percentile60.00% | Participants31 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score81.30% | Percentile56.67% | Participants31 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score79.30% | Percentile53.33% | Participants31 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score79.00% | Percentile50.00% | Participants31 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score78.40% | Percentile46.67% | Participants31 | EvidenceC | Evaluated |
| Rank18 | ModelAC | Score77.80% | Percentile43.33% | Participants31 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score77.40% | Percentile40.00% | Participants31 | EvidenceC | Evaluated |
| Rank20 | ModelAC | Score73.70% | Percentile36.67% | Participants31 | EvidenceC | Evaluated |
| Rank21 | ModelAC | Score73.50% | Percentile33.33% | Participants31 | EvidenceC | Evaluated |
| Rank22 | ModelAC | Score73.20% | Percentile30.00% | Participants31 | EvidenceC | Evaluated |
| Rank23 | ModelLA | Score73.10% | Percentile26.67% | Participants31 | EvidenceC | Evaluated |
| Rank24 | ModelAC | Score71.50% | Percentile23.33% | Participants31 | EvidenceC | Evaluated |
| Rank25 | ModelAC | Score70.90% | Percentile20.00% | Participants31 | EvidenceC | Evaluated |
| Rank26 | ModelAC | Score70.30% | Percentile16.67% | Participants31 | EvidenceC | Evaluated |
| Rank27 | ModelXA | Score68.70% | Percentile13.33% | Participants31 | EvidenceC | Evaluated |
| Rank28 | ModelDE | Score68.40% | Percentile10.00% | Participants31 | EvidenceC | Evaluated |
| Rank29 | ModelDE | Score65.40% | Percentile6.67% | Participants31 | EvidenceC | Evaluated |
| Rank30 | ModelDE | Score64.20% | Percentile3.33% | Participants31 | EvidenceC | Evaluated |
RealWorldQA Highlights
The leading models and scores on this benchmark.
RealWorldQA Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for RealWorldQA
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is RealWorldQA?
What RealWorldQA measures and how its scores work.
RealWorldQA is a benchmark released by xAI as part of the Grok-1.5 Vision preview, consisting of images accompanied by questions and easily verifiable answers.
It measures basic real-world spatial understanding, including the ability to understand natural scenes and spatial relationships in everyday visual contexts, using Score reported as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- RealWorldQA
- Modality
- multimodal
- Primary category
- spatial reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about RealWorldQA.
Which model scores highest on RealWorldQA?
Qwen3.8-Flash-Next is currently ranked first with 88.50%.
What are the top three models on RealWorldQA?
The current leaders are Qwen3.8-Flash-Next (88.50%), Qwen3.8 Flash (88.50%), and Qwen3.8 Max (88.00%).
Which RealWorldQA model has the lowest official input price?
Qwen2.5-Omni-7B has the lowest matched official input price at $0.10 input / $0.40 output per 1M tokens.
Which models are fastest among RealWorldQA results?
The fastest matched records are Qwen3.8 Max (60.87 tok/s via DeepInfra) and Qwen3 VL 4B Thinking (8.77 tok/s via DeepInfra).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.