Models/Blackfrost/BF-GLM52-NVFP4-001

GLM-5.2 — De-Risked NVFP4

Public repositoryFree · Ungated

Z.AI · GLM-5.2 · NVFP4 · free public download

README.mdBF-GLM52-NVFP4-001 · free public access
text-generationnvfp4mixture-of-expertspublicfree

GLM-5.2 — De-Risked NVFP4

The full 256-expert GLM-5.2 research build in NVFP4.

Download free ↓
Free and ungated

Download the public weights directly from Hugging Face. No payment, account, or access request is required. Public GLM-5.2 and GLM-5.3 downloads remain available independently of the paid store.

License terms

The MIT license requires the Zhipu AI copyright and permission notice to accompany every copy or substantial portion of the upstream material. The repository’s existing license terms and notices still apply. Read the upstream license ↗

Release identity and selection guidance

Edition roleResearch quant
Weight package~451 GB · NVFP4 · 159 safetensors shards
Intended fitFootprint-sensitive evaluation of the full 256-expert GLM-5.2 topology.
Choose this edition whenChoose this edition for the smallest listed full-256-expert GLM-5.2 package; validate NVFP4 support before committing to a serving design.
Deployment boundaryNo serving runtime or hardware target is documented for this research build, and the purchase does not include deployment engineering.
Verified repository revision518657480744 ↗

Technical specifications

Base modelzai-org/GLM-5.2
ArchitectureGlmMoeDsaForCausalLM
Upstream licenseMIT ↗
Release identifierBF-GLM52-NVFP4-001
Parameters~753B deployed · 256 routed experts
PrecisionNVFP4 · microscaled 4-bit floating point
Footprint~451 GB · 159 safetensors shards
ContextUp to 1M positions

What is available

The model weights and supporting files published in the named repository, including configuration, tokenizer files when present, and documentation. Review the model card for its current file inventory and usage terms.

Compute, hosting, installation, custom engineering, and ongoing support are separately scoped. Explore hosting or custom services.

Deployment notice

Validate model behavior, runtime compatibility, infrastructure capacity, and operational controls against your workload. Listed configurations are references, not performance guarantees.