GLM-4.6V

GLM-4.6V ranks first on OmniSpatial.

Z.ai Released 8 Dec 2025 Family GLM-V The lab's model card

Benchmarks with a score

Ranked first, out of five or more models

In the top three

Z.ai Dec 2025

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
LMArena (Elo)	1,378	#154 of 333	Official leaderboard
MMBench-CN v1.1	88.2%	#3 of 5	Model card
HallusionBench	62.3%	#16 of 23	Model card
MMLongBench	64.1%	#4 of 4	Model card

Benchmark	Score		Rank	Source
Flame-React-Eval	86.3%		#1 of 3	Model card

Benchmark	Score	Rank	Source
SnakeBench	26	#61 of 183	Official leaderboard
OSWorld	37.2%	#7 of 12	Model card
WebVoyager	81%	#7 of 11	Model card
AndroidWorld	57%	#16 of 18	Model card
GameWorld	30.8%	#12 of 12	Official leaderboard
MM-BrowseComp	7.60%	#5 of 5	Model card

Benchmark	Score		Rank	Source
CADGenBench	18.3%		#10 of 11	Official leaderboard

Benchmark	Score	Rank	Source
CVBench	87.6%	#2 of 11	Model card
OmniSpatial	52%	#1 of 5	Model card
MathVista	85.2%	#13 of 64	Model card
OCRBench v2	65.1%	#6 of 20	Model card
WebQuest	69.3%	#1 of 3	Model card
MMMU	76%	#34 of 97	Model card
BLINK	65.5%	#11 of 28	Model card
MuirBench	77.1%	#5 of 12	Model card
MMStar	75.9%	#13 of 30	Model card
AI2D	88.8%	#20 of 44	Model card
MMBench-EN v1.1	88.8%	#8 of 17	Model card
Video-MME	81.8%	#20 of 42	Model card
ZeroBench	25.8%	#12 of 25	Model card
MathVerse	75.4%	#6 of 12	Model card
MADQA	66.1%	#9 of 16	Official leaderboard
OCRBench	86.5%	#16 of 28	Model card
TreeBench	51.4%	#4 of 7	Model card
GeoBench	71%	#28 of 48	Official leaderboard
MVBench	74.9%	#9 of 15	Model card
ChartMuseum	58.4%	#3 of 5	Model card
ChartQA Pro	65.5%	#5 of 8	Model card
We-Math	69.8%	#7 of 11	Model card
MMMU-Pro	66%	#48 of 72	Model card
Ref-L4	88.9%	#2 of 3	Model card
RefCOCO / RefCOCO+ / RefCOCOg	88.6%	#13 of 19	Model card
Vision Arena	1,161	#94 of 136	Official leaderboard
MMLongBench-Doc	54.9%	#9 of 13	Model card
DynaMath	54.5%	#12 of 17	Model card
MATH-Vision (MathVision)	63.5%	#30 of 42	Model card
MMVU	68.4%	#15 of 20	Model card
LogicVista	62.1%	#9 of 12	Model card
MotionBench	63.6%	#7 of 9	Model card
Video-MMMU	74.7%	#24 of 30	Model card
BabyVision	17.6%	#22 of 27	Official leaderboard
CharXiv	63.2%	#55 of 67	Model card
All-Angles Bench	56.2%	#5 of 6	Model card
LVBench	59.5%	#21 of 25	Model card
ERQA	47.8%	#24 of 26	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. 11 of them aren't in the launch post.