Step-R1-V-Mini: benchmark results

Provider: StepFun. Released 2025-04-08. Access: API.

Unified ELO 1683 ± 30, rank #360 of 2656 rated models, from 22 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MathVision56.6Overall Accuracy (%)70.6
RBench-V13.2Overall Score (%)63
AGI-Eval Community - Interaction (Chinese)78.27Accuracy (%)57.2
AGI-Eval Community - Interaction80.9Accuracy (%)55.7
AGI-Eval Community - General Reasoning95.3Accuracy (%)37.9
AGI-Eval Community - Interaction (English)88.29Accuracy (%)35.5
AGI-Eval Community - Subject Knowledge80.45Accuracy (%)35
AGI-Eval Community - Learning (English)85.76Accuracy (%)31.5
AGI-Eval Community - Mathematical Reasoning60.43Accuracy (%)28.6
AGI-Eval Community - Objective Accuracy (English)83.39Accuracy (%)28.3
AGI-Eval Community - Cognition (Chinese)69.98Accuracy (%)26.8
AGI-Eval Community - Cognition72.39Accuracy (%)25.7

Interactive version: theaggregate.ai/model?slug=step-r1-v-mini · How It Works · Data refreshed daily, snapshot 2026-09-19.