VECTOR WIREAI INTELLIGENCE
UTC
Refresh Models Deals Regulatory Sources

InternVL3.5-8B-Instruct vs MiMo Embodied 7B

10 SHARED BENCHMARKS

Across 10 shared benchmarks, InternVL3.5-8B-Instruct scores higher on 6 and MiMo Embodied 7B on 4. The widest gap is MMStar, where InternVL3.5-8B-Instruct scores 64.1 against 22.4.

OPENGVLABVSXIAOMI10 SHARED64 HEAD-TO-HEAD
CharXiv41.757.5
CountQA20.922.6
Ego3D RMSE ↓239.8
ERQA4246.8
MMMU-Pro std46.427.4
MMMU-Pro vis42.328.1
MMStar64.122.4
OCRBench83.278.8
RealWorldQA66.928.5
WaymoQA all58.169.6

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.