InternVL3-38B — benchmark results

Shanghai AI Lab's open InternVL3 38B vision-language model. Provider: Shanghai AI Lab. Released 2025-04-11. Access: Open.

Unified ELO 1613 ± 5, rank #449 of 1776 rated models, from 1034 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MEGA-Bench Task - ARC AGI14.3Task Score (%)100
MEGA-Bench Task - Adapted CVBench Count78.6Task Score (%)100
MEGA-Bench Task - Code Visualization Output Understanding70Task Score (%)100
MEGA-Bench Task - Counting Single Image85.7Task Score (%)100
MEGA-Bench Task - Entertainment Web Game Style92.9Task Score (%)100
MEGA-Bench Task - Functionality Matching In Different Objects78.6Task Score (%)100
MEGA-Bench Task - Signage Navigation92.9Task Score (%)100
MEGA-Bench Task - Star Object Interaction Video75Task Score (%)100
MEGA-Bench Task - Video Motion Matching 3D Real80Task Score (%)100
MEGA-Bench Task - Video Motion Matching Real 3D85.7Task Score (%)100
MEGA-Bench Task - Visual Correspondence In Two Images92.9Task Score (%)100
OpenVLM MMT-Bench - Clothes Keypoint Detection95Score (%)100

Interactive version: theaggregate.ai/model?slug=internvl3-38b · How the rankings work · Data refreshed daily, snapshot 2026-07-22.