spatial reasoning benchmark
RealWorldQA Leaderboard
RealWorldQA is a benchmark designed to evaluate basic real-world spatial understanding capabilities of multimodal models. The initial release consists of over 700 anonymized images taken from vehicles and other real-world scenarios, each accompanied by a question and easily verifiable answer. Released by xAI as part of their Grok-1.5 Vision preview to test models' ability to understand natural scenes and spatial relationships in everyday visual contexts.
Updated Aug 17, 2026
Models29
Model coverage29
MetricScore
EvidenceB
RealWorldQA Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score88.0% | Percentile100.0% | Participants29 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score86.9% | Percentile96.4% | Participants29 | EvidenceC | Evaluated |
| Rank03 | ModelBY | Score86.7% | Percentile92.9% | Participants29 | EvidenceC | Evaluated |
| Rank04 | ModelBY | Score86.3% | Percentile89.3% | Participants29 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score85.9% | Percentile85.7% | Participants29 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score85.4% | Percentile82.1% | Participants29 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score85.3% | Percentile78.6% | Participants29 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score85.1% | Percentile75.0% | Participants29 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score84.1% | Percentile71.4% | Participants29 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score84.1% | Percentile67.9% | Participants29 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score83.7% | Percentile64.3% | Participants29 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score81.3% | Percentile60.7% | Participants29 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score79.3% | Percentile57.1% | Participants29 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score79.0% | Percentile53.6% | Participants29 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score78.4% | Percentile50.0% | Participants29 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score77.8% | Percentile46.4% | Participants29 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score77.4% | Percentile42.9% | Participants29 | EvidenceC | Evaluated |
| Rank18 | ModelAC | Score73.7% | Percentile39.3% | Participants29 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score73.5% | Percentile35.7% | Participants29 | EvidenceC | Evaluated |
| Rank20 | ModelAC | Score73.2% | Percentile32.1% | Participants29 | EvidenceC | Evaluated |
| Rank21 | ModelLA | Score73.1% | Percentile28.6% | Participants29 | EvidenceC | Evaluated |
| Rank22 | ModelAC | Score71.5% | Percentile25.0% | Participants29 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score70.9% | Percentile21.4% | Participants29 | EvidenceC | Evaluated |
| Rank24 | ModelAC | Score70.3% | Percentile17.9% | Participants29 | EvidenceC | Evaluated |
| Rank25 | ModelXA | Score68.7% | Percentile14.3% | Participants29 | EvidenceC | Evaluated |
| Rank26 | ModelDE | Score68.4% | Percentile10.7% | Participants29 | EvidenceC | Evaluated |
| Rank27 | ModelDE | Score65.4% | Percentile7.1% | Participants29 | EvidenceC | Evaluated |
| Rank28 | ModelDE | Score64.2% | Percentile3.6% | Participants29 | EvidenceC | Evaluated |
| Rank29 | ModelCO | Score62.2% | Percentile0.0% | Participants29 | 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.
Ranking basisThis realworldqa AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01ACQwen3.8 MaxAlibaba Cloud / Qwen Team
- Score
- 88.0%
- Price
- $2.0 input / $6.0 output per 1M tokens
Strengths
- Ranks #1 of 29 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures RealWorldQA, not total model capability
- 02ACQwen3.7-PlusAlibaba Cloud / Qwen Team
- Score
- 86.9%
- Price
- $0.50 input / $3.0 output per 1M tokens
Strengths
- Ranks #2 of 29 compared models
- 96th percentile on this benchmark
- C evidence result
Considerations
- This result measures RealWorldQA, not total model capability
- 03BYSeed 2.1 ProByteDance
- Score
- 86.7%
Strengths
- Ranks #3 of 29 compared models
- 93th percentile on this benchmark
- C evidence result
Considerations
- This result measures RealWorldQA, not total model capability
- 04BYSeed 2.1 TurboByteDance
- Score
- 86.3%
Strengths
- Ranks #4 of 29 compared models
- 89th percentile on this benchmark
- C evidence result
Considerations
- This result measures RealWorldQA, not total model capability
- 05ACQwen3.8-27BAlibaba Cloud / Qwen Team
- Score
- 85.9%
Strengths
- Ranks #5 of 29 compared models
- 86th percentile on this benchmark
- C evidence result
Considerations
- This result measures RealWorldQA, not total model capability
Selection summary
Best AI Models for RealWorldQA
Qwen3.8 Max currently leads RealWorldQA with 88.0%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is RealWorldQA?
What RealWorldQA measures and how its scores work.
RealWorldQA is a benchmark designed to evaluate basic real-world spatial understanding capabilities of multimodal models. The initial release consists of over 700 anonymized images taken from vehicles and other real-world scenarios, each accompanied by a question and easily verifiable answer. Released by xAI as part of their Grok-1.5 Vision preview to test models' ability to understand natural scenes and spatial relationships in everyday visual contexts.
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
- Yes
- Evaluation key
- realworldqa|llm-stats-current
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 Max is currently ranked first with 88.0%.
What does RealWorldQA measure?
RealWorldQA is a benchmark designed to evaluate basic real-world spatial understanding capabilities of multimodal models. The initial release consists of over 700 anonymized images taken from vehicles and other real-world scenarios, each accompanied by a question and easily verifiable answer. Released by xAI as part of their Grok-1.5 Vision preview to test models' ability to understand natural scenes and spatial relationships in everyday visual contexts.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
29 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.