A general visual question-answering benchmark used in provider tables for real-image reasoning quality.
BenchLM is tracking MStar in the local dataset, but exact-source verification records for these rows are still being attached. To avoid a blank benchmark page, BenchLM shows the current tracked rows below as a display-only reference table.
These tracked rows are useful for inspection and spot-checking, but until exact-source attachments are completed they should not be treated as fully verified public benchmark rows.
BenchLM mirrors the published tracked score view for MStar. Qwen3.5 397B leads the public snapshot at 83.8% , followed by Qwen3.6 Plus (83.3%) and Gemini 3 Pro (83.1%). BenchLM does not use these results to rank models overall.
Qwen3.5 397B
Alibaba
qwen3-5-397b
Qwen3.6 Plus
Alibaba
qwen3-6-plus
Gemini 3 Pro
gemini-3-pro
The published MStar snapshot is tightly clustered at the top: Qwen3.5 397B sits at 83.8%, while the third row is only 0.7 points behind. The broader top-10 spread is 10.6 points, so the benchmark still separates strong models even when the leaders cluster.
6 models have been evaluated on MStar. The benchmark falls in the Multimodal & Grounded category. This category carries a 12% weight in BenchLM.ai's overall scoring system. MStar is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
Year
2026
Tasks
Real-image visual QA
Format
Image-grounded QA
Difficulty
General visual reasoning
MStar sits between broad multimodal reasoning and grounded VQA. It is useful for checking whether a model can answer real-image questions without the stronger domain structure of office or academic benchmarks.
Version
MStar 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
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.
A general visual question-answering benchmark used in provider tables for real-image reasoning quality.
Qwen3.5 397B currently leads the published MStar snapshot with a tracked score of 83.8%. BenchLM shows this benchmark for display only and does not use it in overall rankings.
6 AI models are included in BenchLM's mirrored MStar snapshot, based on the public leaderboard captured on April 10, 2026.
For engineers, researchers, and the plain curious — a weekly brief on new models, ranking shifts, and pricing changes.
Free. No spam. Unsubscribe anytime.