reasoning benchmark
ERQA Leaderboard
Embodied Reasoning Question Answering benchmark consisting of 400 multiple-choice visual questions across spatial reasoning, trajectory reasoning, action reasoning, state estimation, and multi-view reasoning for evaluating AI capabilities in physical world interactions
Updated Aug 17, 2026
Models24
Model coverage24
MetricScore
EvidenceC
ERQA Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score77.8% | Percentile100.0% | Participants24 | EvidenceC | Evaluated |
| Rank02 | ModelBY | Score72.0% | Percentile95.7% | Participants24 | EvidenceC | Evaluated |
| Rank03 | ModelBY | Score71.3% | Percentile91.3% | Participants24 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score69.8% | Percentile87.0% | Participants24 | EvidenceC | Evaluated |
| Rank05 | ModelOP | Score65.7% | Percentile82.6% | Participants24 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score65.7% | Percentile78.3% | Participants24 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score65.5% | Percentile73.9% | Participants24 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score64.8% | Percentile69.6% | Participants24 | EvidenceC | Evaluated |
| Rank09 | ModelME | Score64.7% | Percentile65.2% | Participants24 | EvidenceC | Evaluated |
| Rank10 | ModelOP | Score64.0% | Percentile60.9% | Participants24 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score62.5% | Percentile56.5% | Participants24 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score62.0% | Percentile52.2% | Participants24 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score60.5% | Percentile47.8% | Participants24 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score52.5% | Percentile43.5% | Participants24 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score52.3% | Percentile39.1% | Participants24 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score51.3% | Percentile34.8% | Participants24 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score48.8% | Percentile30.4% | Participants24 | EvidenceC | Evaluated |
| Rank18 | ModelAC | Score47.3% | Percentile26.1% | Participants24 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score46.8% | Percentile21.7% | Participants24 | EvidenceC | Evaluated |
| Rank20 | ModelAC | Score45.8% | Percentile17.4% | Participants24 | EvidenceC | Evaluated |
| Rank21 | ModelAC | Score45.3% | Percentile13.0% | Participants24 | EvidenceC | Evaluated |
| Rank22 | ModelAC | Score43.0% | Percentile8.7% | Participants24 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score41.3% | Percentile4.3% | Participants24 | EvidenceC | Evaluated |
| Rank24 | ModelOP | Score35.2% | Percentile0.0% | Participants24 | EvidenceC | Evaluated |
ERQA Highlights
The leading models and scores on this benchmark.
ERQA Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for ERQA
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis erqa 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
- 77.8%
- Price
- $2.0 input / $6.0 output per 1M tokens
Strengths
- Ranks #1 of 24 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures ERQA, not total model capability
- 02BYSeed 2.1 ProByteDance
- Score
- 72.0%
Strengths
- Ranks #2 of 24 compared models
- 96th percentile on this benchmark
- C evidence result
Considerations
- This result measures ERQA, not total model capability
- 03BYSeed 2.1 TurboByteDance
- Score
- 71.3%
Strengths
- Ranks #3 of 24 compared models
- 91th percentile on this benchmark
- C evidence result
Considerations
- This result measures ERQA, not total model capability
- 04ACQwen3.7-PlusAlibaba Cloud / Qwen Team
- Score
- 69.8%
- Price
- $0.50 input / $3.0 output per 1M tokens
Strengths
- Ranks #4 of 24 compared models
- 87th percentile on this benchmark
- C evidence result
Considerations
- This result measures ERQA, not total model capability
- 05OPGPT-5OpenAI
- Score
- 65.7%
- Price
- $1.3 input / $10 output per 1M tokens
- Speed
- Up to 100 tok/s via OpenAI
Strengths
- Ranks #5 of 24 compared models
- 83th percentile on this benchmark
- C evidence result
Considerations
- This result measures ERQA, not total model capability
Selection summary
Best AI Models for ERQA
Qwen3.8 Max currently leads ERQA with 77.8%. 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 ERQA?
What ERQA measures and how its scores work.
Embodied Reasoning Question Answering benchmark consisting of 400 multiple-choice visual questions across spatial reasoning, trajectory reasoning, action reasoning, state estimation, and multi-view reasoning for evaluating AI capabilities in physical world interactions
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of C.
- Family
- ERQA
- Modality
- multimodal
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- erqa|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about ERQA.
Which model scores highest on ERQA?
Qwen3.8 Max is currently ranked first with 77.8%.
What does ERQA measure?
Embodied Reasoning Question Answering benchmark consisting of 400 multiple-choice visual questions across spatial reasoning, trajectory reasoning, action reasoning, state estimation, and multi-view reasoning for evaluating AI capabilities in physical world interactions
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
24 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.