Kimi-K2.6
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
• Data + Expert ParallelFits inside one node's interconnect. single_node_dep
• Multi-Node Tensor ParallelSpans nodes. multi_node_tp
• Multi-Node TP + Pipeline ParallelSpans nodes. multi_node_tp_pp
• Multi-Node Data + Expert ParallelSpans nodes. multi_node_dep
• Multi-Node Tensor + Expert ParallelSpans nodes. multi_node_tep
• Prefill/Decode DisaggregationSpans nodes. pd_cluster

• B300 · 8 GPUs
• GB200 NVL4 · 4 GPUs
• H200 · 8 GPUs
• MI300X · 8 GPUs
• MI325X · 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 moonshotai/Kimi-K2.6 \
  --trust-remote-code \
  --tool-call-parser kimi_k2 \
  --enable-auto-tool-choice \
  --reasoning-parser kimi_k2 \
  --speculative-config '{"model":"lightseekorg/kimi-k2.6-eagle3-mla","method":"eagle3","num_speculative_tokens":3}' \
  --mm-encoder-tp-mode data

Features

• Tool callingOn by defaultKimi K2 tool-call parser with automatic tool choice
• ReasoningOn by defaultKimi K2 reasoning parser for extracting chain-of-thought content
• Speculative decodingOn by defaultEagle3 speculative decoding for accelerated inference
• Text-only modeOpt-inSkip loading the vision encoder for text-only workloads — frees VRAM for KV cache. Mutually exclusive with encoder_parallel.
• Encoder data parallelOn by defaultRun the vision encoder in data-parallel mode — avoids TP comm overhead on the small encoder. Mutually exclusive with text_only.

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:moonshotai/Kimi-K2.6"

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 Kimi-K2.6 need?The vLLM recipe publishes a minimum of 600 GB across the whole replica for its NVFP4 variant; INT4 714 GB. That is vLLM's own measurement, not a per-GPU figure.
• Which parallelism layouts does the Kimi-K2.6 recipe support?Tensor Parallel, Tensor + Expert Parallel, Data + Expert Parallel, Multi-Node Tensor Parallel, Multi-Node TP + Pipeline Parallel, Multi-Node Data + Expert Parallel, Multi-Node Tensor + Expert Parallel and Prefill/Decode Disaggregation.
• Can Wingsmith serve Kimi-K2.6?Wingsmith serves models through NVIDIA Dynamo on clusters with NVIDIA GPUs. Register moonshotai/Kimi-K2.6 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.
• Does NVIDIA publish a Dynamo recipe for Kimi-K2.6?Yes: 4 deployment entries for vLLM, aggregated.

From the vLLM recipe for moonshotai/Kimi-K2.6 ↗ (snapshot at 4bcc189) and NVIDIA Dynamo's recipes at 165d854, Apache 2.0.