GLM-5.3 5.3

Z.ai · released Aug 14, 2026

Unknown

Openness taxonomy

access, licence, and disclosure — never a single open/closed binary

Weights availability: Weights expected in about two weeks; not yet available at publication
Licence: unknown
Licence category: unknown
Commercial use allowed: unknown
Derivatives allowed: unknown
Redistribution allowed: unknown
Field-of-use restrictions: unknown
Training code availability: unknown
Inference code availability: unknown
Architecture disclosure: Base model reuse and scale are disclosed
Training data disclosure: unknown

openness confidence

Weights expected in about two weeks

Deployability

raw values as extracted — free text, never inferred from branding

Self-hosting: unknown
On-premise: unknown
Edge deployability: unknown
API availability: unknown
Regions available: unknown

Benchmarks

as reported — vendor-supplied unless otherwise noted

Terminal-Bench 3.0 28.3, up from 4.6DeepSWE v1.1 66.9, up from 46.2CyberGym 84.5%ExploitBench 54.4%, more than double the previous result

National-security pathway & evidence

from the assessed source item, when one is linked

Impact: highConfidence: mediumNat-sec: indirectindependent reporting

Domains: cyber, ai_system_security

Pathway: If independently validated, stronger long-horizon coding and exploit-oriented task performance could improve the ability of users to automate portions of cyber research and operations.

Evidence limitations: The item is independent reporting based on reported Z.ai results; no independent benchmark reproduction is provided, and weights are not yet available for testing.

Z.ai Ships GLM-5.3 Without Retraining the Base Model: Better at Complex Coding and Long-Horizon Tasks

Linked research papers

from paper_model_links

Who Verifies the Benchmark? Decentralizing Trust in Large Language Model Evaluationevaluates · 98% confidence