Will it last?
This page is a projection of the one entry record, the Transparency factor that Support covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.
- AssessUse & modify + Transparency
- ImplementData control + Reliability
- UseReliability
- SupportTransparency
Common problems & fixes
The recurring issues are 1M-context memory blow-ups, older chat-template / tool-call conventions that differ from the current GLM line, and quant-quality trade-offs. The list is not exhaustive; the community around this widely-used 2024 model is the practical first stop.
Versions, changelog & cadence
The original GLM-4-9B-Chat / -Chat-1M (and the glm-4-9b base) are on the THUDM organisation. For new work they are succeeded by GLM-4-9B-0414 and the 4.5/4.6 line (the glm entry). The Hugging Face revision hash is the changelog anchor.
Security / vulnerability disclosure
A licence contact exists (license@zhipuai.cn), but no formal vulnerability-disclosure / security policy exists for the weights. Factor this into your own incident-response plan.
Community & support channels
The THUDM Hugging Face organisation and GitHub, plus a community around this widely-used 2024
small model.
Deprecation / end-of-life policy
No published deprecation or end-of-life policy. The model is effectively replaced by the current GLM line but remains downloadable, with no documented sunset commitment; recorded as a gap.
Tracked known issues
The standing issues are the revocable, registration-gated licence, dated capability, China-aligned alignment, and the absence of a first-party guard model. See Assess for the detail.
How this scores
The ownership factor this domain covers, drawn from the one entry record.
TransparencyDo you know what it is: weights, training, behaviour, and legible terms?
ModerateWeights are inspectable and there is a model card, but training data and code are closed and the model carries China-aligned alignment you cannot inspect - open weights but closed process, so moderate.
Sources
The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.