VECTOR WIREAI INTELLIGENCE
UTC
Refresh Models Deals Regulatory Sources

InternVL-3-78B vs Qwen2.5 VL 3B

8 SHARED BENCHMARKS

Across 8 shared benchmarks, InternVL-3-78B scores higher on 6 and Qwen2.5 VL 3B on 2. The widest gap is OCRVQA_TEST, where Qwen2.5 VL 3B scores 69.2 against 35.6.

OPENGVLABVSALIBABA8 SHARED62 HEAD-TO-HEAD
AI2D83.577.1
BLINK51.949
DocVQA-val83.892.7
MMStar66.156.1
OCRVQA_TEST35.669.2
POPE88.986.2
RealWorldQA74.365.2
TextVQA-val83.579.1

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.