Moonshot's 2.8T-parameter multimodal flagship for long-horizon coding, reasoning, and agent workflows.
Use Moonshot's official vLLM or SGLang recipe; a single-node container command is not a production deployment for this model.
Every card opens a setup sequence tailored to this model's runtime and memory floor.
The planning floor exceeds one provider node; choose a cluster-capable provider.
Serverless vLLM for smaller models; dedicated Pods for multi-GPU models Open runbook →The planning floor exceeds one provider node; choose a cluster-capable provider.
A Docker template attached to a matching GPU marketplace offer Open runbook →Requires the provider's multi-node path and the model author's distributed recipe.
An On-Demand Cloud instance, or a 1-Click Cluster for larger deployments Open runbook →Requires the provider's multi-node path and the model author's distributed recipe.
SageMaker HyperPod with a Hugging Face model source and vLLM/SGLang worker Open runbook →Requires the provider's multi-node path and the model author's distributed recipe.
A custom vLLM container in Vertex AI, backed by Artifact Registry Open runbook →Requires the provider's multi-node path and the model author's distributed recipe.
Azure Machine Learning online endpoint with a bring-your-own vLLM/SGLang container Open runbook →Requires the provider's multi-node path and the model author's distributed recipe.
OCI Generative AI imported models when supported; OKE vLLM otherwise Open runbook →Start at 32,768 tokens, confirm all GPUs are visible, load the official chat template, and send a short request. Increase context and concurrency only after recording free VRAM.
Rank comes from observed OpenRouter ordering. Parameters, context, license, and runtime notes come from the linked model card. The 16×192GB hardware figure is a planning floor with runtime and KV-cache headroom.
Open a terminal in the repository where you want the deployment files,
start claude or codex, then paste this prompt.
It asks the agent to verify sources and stop before it creates billable infrastructure.
Prepare a reproducible deployment project for Kimi K3 (moonshotai/Kimi-K3); ask me to choose a provider before writing provider-specific infrastructure. Use this guide as the starting context: https://www.getflops.ai/models/kimi-k3. Use a planning floor of 16x 192GB GPUs, 3800GB storage, sglang latest, and an initial context limit of 32768 tokens. Open every linked primary source and flag any mismatch instead of guessing. Create a deployment folder containing README.md, .env.example with no secrets, a pinned start script or infrastructure manifest, and smoke-test.sh. Make the endpoint OpenAI-compatible where the runtime supports it. Run local/static validation, estimate the billable resources, and stop before provisioning paid infrastructure until I approve.Research checked 2026-07-27. Provider inventory and inference-engine support can change faster than this page.