Benchmark Almanac

LLaVA-OneVision 7B

LLaVA-OneVision 7B places in the top three on OVBench.

ByteDance Seed Released 6 Aug 2024 Family LLaVA-OneVision The lab's model card
23
Benchmarks with a score
0
Ranked first, out of five or more models
1
In the top three
ByteDance Seed Aug 2024

LLaVA-OneVision 7B

Open weights Small

No Epoch Capabilities Index score yet
  • OCRBench v2 33.7% #50 /65
  • VisualPuzzles 29.4% #23 /29
  • HumanEval-V 1.60% #17 /24

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
MMBench-CN v1.1 80.9% #41 of 178 Official leaderboard
Creation-MMBench -43.49 #17 of 19 Official leaderboard

Coding

BenchmarkScoreRankSource
HumanEval-V 1.60% #17 of 24 Official leaderboard

Vision, audio and robotics

BenchmarkScoreRankSource
MMBench 83.2% #31 of 155 Official leaderboard
MMBench-EN v1.1 80.9% #46 of 178 Official leaderboard
A-Bench 75.2% #32 of 116 Official leaderboard
MMStar 61.9% #48 of 173 Official leaderboard
CCBench (MMBench Chinese Culture) 54.9% #52 of 152 Official leaderboard
OVBench 49.5% #3 of 7 Official leaderboard
TOMATO 25.5% #14 of 22 Official leaderboard
ERQA 34.4% #11 of 17 Official leaderboard
KRETA 54% #11 of 17 Official leaderboard
MMMU 48.8% #89 of 136 Official leaderboard
OCRBench v2 33.7% #50 of 65 Official leaderboard
MEGA-Bench 23.9% #17 of 22 Official leaderboard
MMVU 37.9% #22 of 28 Official leaderboard
CharXiv 23.6% #61 of 77 Official leaderboard
VisualPuzzles 29.4% #23 of 29 Official leaderboard
MMMU-Pro 24.1% #48 of 57 Official leaderboard
Video-MMMU 33.9% #20 of 23 Official leaderboard
Point-Bench 6.32% #17 of 17 Official leaderboard
MME-CoT 30.9% #11 of 11 Official leaderboard

Other

BenchmarkScoreRankSource
OpenVLM Leaderboard 60.2 #82 of 173 Official leaderboard