zai-org/GLM-4.5V-FP8

Weights: 110.0 GB · safetensors
Source: huggingface page
Commit: 3ca028e

Evaluated across 12 benchmarks. Ranks in the top 3 on 0 of 12. Strongest showing on MMLU (81.0% acc, #6 of 28). Weakest on MBPP (0.0% pass@1, #28 of 28).

Benchmark results

Benchmark	Metric	Score	Rank	Actions
MMLU	acc	81.0%±0.3%	#6 of 28	View run →
MGSM	exact_match	73.0%±0.7%	#8 of 28	View run →
IFEval	prompt_level_strict_acc	69.9%±2.0%	#21 of 28	View run →
GPQA Diamond	acc	28.3%±3.2%	#19 of 28	View run →
LongBench	aggregate	27.3%±0.4%	#7 of 17	View run →
GPQA Main	acc	24.3%±2.0%	#27 of 28	View run →
GPQA Extended	acc	23.8%±1.8%	#25 of 28	View run →
MMLU-Pro	exact_match	6.0%±0.2%	#28 of 28	View run →
GSM8K	exact_match	3.5%±0.5%	#27 of 28	View run →
BBH	exact_match	2.7%±0.2%	#20 of 28	View run →
EQ-Bench	eqbench	1.1±0.8	#27 of 28	View run →
MBPP	pass@1	0.0%±0.0%	#28 of 28	View run →

MMLU#6 of 28
acc
81.0%±0.3%
View run →
MGSM#8 of 28
exact_match
73.0%±0.7%
View run →
IFEval#21 of 28
prompt_level_strict_acc
69.9%±2.0%
View run →
GPQA Diamond#19 of 28
acc
28.3%±3.2%
View run →
LongBench#7 of 17
aggregate
27.3%±0.4%
View run →
GPQA Main#27 of 28
acc
24.3%±2.0%
View run →
GPQA Extended#25 of 28
acc
23.8%±1.8%
View run →
MMLU-Pro#28 of 28
exact_match
6.0%±0.2%
View run →
GSM8K#27 of 28
exact_match
3.5%±0.5%
View run →
BBH#20 of 28
exact_match
2.7%±0.2%
View run →
EQ-Bench#27 of 28
eqbench
1.1±0.8
View run →
MBPP#28 of 28
pass@1
0.0%±0.0%
View run →

last evaluated:6 months ago

How to cite

Citation

FrozeBench. "zai-org/GLM-4.5V-FP8." https://frozebench.com/models/zai-org-glm-4-5v-fp8. Retrieved 2026-06-04.

BibTeX

@misc{frozebench_zai_org_GLM_4_5V_FP8,
  title = {zai-org/GLM-4.5V-FP8},
  howpublished = {\url{https://frozebench.com/models/zai-org-glm-4-5v-fp8}},
  year = {2026},
  note = {FrozeBench. Retrieved 2026-06-04.}
}

URL

https://frozebench.com/models/zai-org-glm-4-5v-fp8