GLM-4.7

Training compute

4.4×10²⁴ FLOP

Parameters

358B

Published

Dec 22, 2025

GLM-4.7 is an AI model developed by Z.ai (Zhipu AI) (China), first published in December 2025. It works in the language domain, on tasks such as language modeling/generation and code generation.

Training it took an estimated 4.4×10²⁴ FLOP of compute. The model has 358,000,000,000 parameters.

Access: Open weights (unrestricted). Its weights are openly available. It is built on top of GLM-4.5. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Z.ai (Zhipu AI)
Country of organization: China
Domain: Language
Task: Language modeling/generation, Code generation
Training compute: 4.4×10²⁴ FLOP
Parameters: 358,000,000,000
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Base model: GLM-4.5
Epoch confidence: Likely

Benchmark results

01	Vending-Bench 2	Score	2376.8
02	WebDev Arena	Arena Score	1441
03	ALE Bench	Performance	399.5
04	Epoch Capabilities Index	ECI Score	143.8
05	GPQA Diamond	mean_score	0.83
06	OTIS Mock AIME 2024-2025	mean_score	0.83
07	SimpleBench	Score (AVG@5)	0.48
08	Terminal-Bench	Accuracy mean	0.33
09	SimpleQA Verified	mean_score	0.32
10	Posttrainbench	Average (%)	0.07
11	Proofbench	Accuracy	0.06
12	Chess Puzzles	mean_score	0.06