GLM-5.2
serving recipe.

Floors are the minimum the vLLM recipe publishes across the whole replica, from vLLM's own runs. The control plane compares them with its own sizing from the model config.

• Tensor ParallelFits inside one node's interconnect. single_node_tp
• Tensor + Expert ParallelFits inside one node's interconnect. single_node_tep
• Multi-Node Tensor ParallelSpans nodes. multi_node_tp
• Multi-Node TP + Pipeline ParallelSpans nodes. multi_node_tp_pp
• Multi-Node Tensor + Expert ParallelSpans nodes. multi_node_tep
• Multi-Node Data + Expert ParallelSpans nodes. multi_node_dep
• Prefill/Decode DisaggregationSpans nodes. pd_cluster

• B200 · 8 GPUs
• B300 · 8 GPUs
• MI300X · 8 GPUs
• MI355X · 8 GPUs

Wingsmith runs models through NVIDIA Dynamo, so it serves on NVIDIA GPUs. Other vendors are listed as the recipe states them.

vllm serve zai-org/GLM-5.2-FP8 \
  --kv-cache-dtype fp8_e4m3 \
  --tool-call-parser glm47 \
  --enable-auto-tool-choice \
  --reasoning-parser glm45

Features

• Tool callingOn by defaultGLM-4.7 tool call parser with automatic tool choice
• ReasoningOn by defaultGLM-4.5 reasoning parser — thinking mode enabled by default on requests
• Speculative decodingOpt-inMulti-Token Prediction speculative decoding (up to 7 draft tokens) Modes: MTP, DSpark.

1. Register the weights

   In the console, or through the API. Resolving a slug proposes an artifact and writes nothing.

   curl "http://localhost:8000/v1/artifacts/resolve?source=hf:zai-org/GLM-5.2"

2. Read the draft

   The draft carries this recipe as upstream evidence: the tested flags, the precision and memory floor, the parallelism layouts, and the hardware vLLM verified, with links back to the recipe. Nothing from it is applied without you, and the draft warns when the plane's own sizing and the published floor differ by a quarter or more.

3. Create the workload

   Choose the engine, device shape and scale. The plane picks the clusters, and each cluster's operator renders the deployment as a Dynamo graph.

4. Give it an endpoint

   One model name for clients, routed by the gateway to deployments on every cluster serving it.

Run the control plane locally →

• How much GPU memory does GLM-5.2 need?The vLLM recipe publishes a minimum of 446 GB across the whole replica for its MXFP4 variant; FP8 893 GB, NVFP4 558 GB, BF16 1786 GB. That is vLLM's own measurement, not a per-GPU figure.
• Which parallelism layouts does the GLM-5.2 recipe support?Tensor Parallel, Tensor + Expert Parallel, Multi-Node Tensor Parallel, Multi-Node TP + Pipeline Parallel, Multi-Node Tensor + Expert Parallel, Multi-Node Data + Expert Parallel and Prefill/Decode Disaggregation.
• Can Wingsmith serve GLM-5.2?Wingsmith serves models through NVIDIA Dynamo on clusters with NVIDIA GPUs. Register zai-org/GLM-5.2 and the plane attaches this vLLM recipe to the draft as evidence: its tested flags, precision and memory floor, for you to review rather than applied on your behalf.

From the vLLM recipe for zai-org/GLM-5.2 ↗ (snapshot at 4bcc189), Apache 2.0.