Serving GLM 5.2 with vLLM
GLM 5.2 has 744 billion parameters. vLLM targets production serving, throughput and concurrency. This page gives the memory required and the platforms that qualify.
What is the minimum memory?
GLM 5.2 totals 744 billion parameters. In NVFP4 its weights take about 428 GB including the runtime margin: 744 billion parameters × 0.5 byte (NVFP4) = 372.0 GB of weights; × 1.15 runtime margin = 427.8 GB. In FP8 the footprint doubles, to about 856 GB (744 billion parameters × 1 byte (FP8) = 744.0 GB of weights; × 1.15 runtime margin = 855.6 GB.) The attention cache is not calculated on this page. These models’ architectures (compressed latent, linear or sparse attention, Mamba layers) share no per-token formula; it is measured on the machine, with the target context and concurrency [TO BE MEASURED].
What is vLLM for?
vLLM targets production serving, throughput and concurrency. See the vLLM and GLM 5.2 fact sheets.
On which platforms?
- GLM 5.2 on Mac Studio Ultra
- GLM 5.2 on DGX Station
- GLM 5.2 on RTX PRO 6000 server
- GLM 5.2 on H200 SXM server
- GLM 5.2 on B200 SXM
- GLM 5.2 on B300 SXM
How much memory for GLM 5.2 with vLLM?
About 428 GB in NVFP4 for the weights, excluding the attention cache.
Does vLLM suit GLM 5.2?
vLLM serves GLM 5.2 as soon as the machine offers at least 428 GB of memory, the weight footprint in NVFP4. 7 of the 8 platforms in the catalogue meet that bar.
Method and sources
Memory footprints are calculated, not measured: a reading on real hardware may differ depending on the engine and the exact weight format. Every figure below carries its source and the date of its last check.
- GLM 5.2: 744 billion parameters, 40 billion active, 1,048,576-token context, MIT licence. Hugging Face repository (config.json, API, model card), checked on 2 September 2026. 744 billion announced by Z.AI; the repository holds 753 billion tensor elements in BF16 (1,507 GB).
- Bytes per parameter: NVFP4 0.5 (4 bits per parameter, hence 0.5 byte; 4-bit published weights weigh 0.54 to 0.56 byte per parameter with their scales (DeepSeek V4 Pro 865 GB for 1,599 billion, Kimi K3 1,561 GB for 2,780 billion)); FP8 1 (8 bits per parameter (E4M3), hence 1 byte, block scales not counted). Source: connaissance/faits.yaml, quantifications family, checked on 2026-08-31.
- Runtime margin × 1.15: QDNA operating assumption, not measured: 15% above the weights for activations, buffers and fragmentation.
- “Fits” threshold at 75% of memory: QDNA assumption, not measured, which keeps the remaining quarter for the attention cache and concurrency.