Vision & documents

LVBench leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap
License: https://creativecommons.org/licenses/by-nc/4.0/

As of 2026-10-29, the highest measured score on LVBench is 91.7% by Gemini 4 Argon. 10 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Gemini 4 Argon	91.7%	measured
2	Gemini 3.8 Flash	87.1%	measured
3	Kimi K3.1 Pro	86.8%	estimated ± 8.2 pp, low confidence
4	Claude Mythos 5	86.6%	estimated ± 8.2 pp, low confidence
5	Gemini 3.7 Flash	85.4%	measured
6	Qwen3.8 Max	81.8%	measured
7	Kimi K3	79.4%	estimated ± 8.2 pp, low confidence
8	Qwen3.8-27B	77.3%	estimated ± 8.2 pp, low confidence
9	Qwen3.8-Omni-Flash	76.9%	measured
10	Qwen3.8-Flash-Next	76.6%	measured
11	dots3-note Preview	76.1%	estimated ± 8.2 pp, low confidence
12	Claude Opus 4.7 (Adaptive)	74.0%	estimated ± 8.2 pp, low confidence
13	Claude Opus 4.8	70.6%	estimated ± 8.2 pp, low confidence
14	Inkling	65.3%	estimated ± 8.2 pp, low confidence
15	Inkling-Small	63.7%	estimated ± 8.2 pp, low confidence
16	Claude Sonnet 5	62.8%	estimated ± 8.2 pp, low confidence
17	LFM3-VL-2B	41.2%	measured

LVBench leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General