Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM

ERQA

A grounded visual reasoning benchmark focused on evidence-based question answering over real images.

Data verified 29 confirmed releases in the last 30 daysStart free brief

The public ERQA snapshot ranks Qwen3.8 Max first at 77.8%, ahead of Qwen3.7 Plus (69.8%) and Gemini 3.1 Pro (69.4%) among 9 tested models. We mirror the table as display-only evidence; it does not affect overall rankings.

Benchmark score on ERQA — August 18, 2026

BenchLM mirrors the published score view for ERQA. Qwen3.8 Max leads the public snapshot at 77.8% , followed by Qwen3.7 Plus (69.8%) and Gemini 3.1 Pro (69.4%). We do not use these results to rank models overall.

1Open

Qwen3.8 Max

Alibaba

77.8%
Overall 79.91Context 1M
2Closed

Qwen3.7 Plus

Alibaba

69.8%
Overall 66.02Context 1M
3Closed

Gemini 3.1 Pro

Google

69.4%
Overall 55.98Context 1M
9 modelsMultimodal & GroundedCurrentDisplay onlyUpdated August 18, 2026

Benchmark score table (9 models)

Score
1
Qwen3.8 MaxAlibaba · Open weight
77.8%
2
Qwen3.7 PlusAlibaba · Closed
69.8%
3
Gemini 3.1 ProGoogle · Closed
69.4%
4
Qwen3.8-27BAlibaba · Open weight
65.5%
5
GPT-5.4OpenAI · Closed
65.4%
6
Muse SparkMeta · Closed
64.7%
7
Qwen3.6-27BAlibaba · Open weight
62.5%
8
Grok 4.20xAI · Closed
54.1%
9
Claude Opus 4.6Anthropic · Closed
51.6%

The published ERQA snapshot places Qwen3.8 Max first at 77.8%. The third row is 8.4 points behind. The broader top-10 range is 26.2 points, so the table still separates the published systems.

9 models have been evaluated on ERQA. The benchmark falls in the Multimodal & Grounded category. This category carries a 12% weight in BenchLM.ai's overall scoring system. ERQA is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About ERQA

Year

2026

Tasks

Evidence-based visual QA

Format

Grounded image reasoning

Difficulty

Grounded multimodal reasoning

ERQA is useful as a grounded reasoning check because it emphasizes answer correctness tied to visual evidence rather than fluent but ungrounded descriptions.

Qwen3.6 launch benchmarks

BenchLM freshness & provenance

Version

ERQA 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does ERQA measure?

A grounded visual reasoning benchmark focused on evidence-based question answering over real images.

Which model scores highest on ERQA?

Qwen3.8 Max by Alibaba currently leads with a score of 77.8% on ERQA.

How many models are evaluated on ERQA?

9 AI models have been evaluated on ERQA on BenchLM.

Compare Top Models on ERQA

Qwen3.8 Max vs Qwen3.7 PlusQwen3.7 Plus vs Gemini 3.1 ProGemini 3.1 Pro vs Qwen3.8-27BQwen3.8-27B vs GPT-5.4
Last updated: August 18, 2026 · BenchLM version ERQA 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.