Keye-VL-2.0-30B-A3B

Keye-VL-2.0-30B-A3B ranks first on 2 of the 27 benchmarks it has scores on.

Kuaishou Released 25 May 2026 Family Keye-VL The lab's model card

Benchmarks with a score

Ranked first, out of five or more models

In the top three

Kuaishou May 2026

Open weights Mid-size

No Epoch Capabilities Index score yet

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.

Benchmark	Score	Rank	Source
HallusionBench	73.5%	#1 of 23	Model card
IFEval	92.2%	#16 of 126	Model card
C-Eval	89.2%	#19 of 38	Model card
Humanity's Last Exam	22.9%	#91 of 147	Model card

Benchmark	Score		Rank	Source
AIME 2025	86.7%		#66 of 143	Model card
AIME 2026	93.3%		#19 of 34	Model card

Benchmark	Score	Rank	Source
OJBench	71.5%	#1 of 12	Model card
LiveCodeBench v6	77.1%	#37 of 76	Model card
SWE-bench Verified	62%	#103 of 150	Model card

Benchmark	Score	Rank	Source
tau2-bench	82.6%	#27 of 69	Model card
BFCL V4	65.7%	#12 of 24	Model card
VITA-Bench	33.1%	#9 of 12	Model card

Benchmark	Score	Rank	Source
OmniDocBench v1.5	84.7	#2 of 13	Model card
MathVista	82.9%	#19 of 64	Model card
We-Math	75.5%	#4 of 11	Model card
MMMU	75.3%	#38 of 97	Model card
DynaMath	80.9%	#7 of 17	Model card
MMStar	75.9%	#14 of 30	Model card
PixMo-Count	83.5%	#1 of 2	Model card
LongVideoBench	74.1%	#6 of 11	Model card
AI2D	87.8%	#27 of 44	Model card
MLVU	82.8%	#10 of 16	Model card
Video-MME	78.3%	#27 of 42	Model card
OCRBench	85.7%	#19 of 28	Model card
Video-MMMU	80%	#22 of 30	Model card
RefCOCO / RefCOCO+ / RefCOCOg	88.1%	#14 of 19	Model card
EmbSpatialBench	83.2%	#8 of 10	Model card
VideoMME v2 (with subtitles)	42.4%	#5 of 6	Model card

The benchmarks the lab chose to headline, with the scores it gave.

Every result in the longer documents the lab published with it.