How to Run Kimi K2 on Ollama: Cloud Setup Guide (2026)

What Kimi K2 Is and Why Ollama Runs It in the Cloud

Kimi K2 is an open-weight large language model from Moonshot AI, a Beijing-based AI lab. The original Kimi K2 (released in 2025) uses a mixture-of-experts architecture with 1 trillion total parameters and 32 billion active per token, pretrained on 15.5 trillion tokens. Moonshot released the weights under a Modified MIT License on GitHub at moonshotai/Kimi-K2. The model is built for agentic work: long multi-step coding sessions, tool use, and autonomous task execution rather than single-turn chat.

The reason Kimi K2 only shows up on Ollama as a cloud model comes down to size. Even at the most aggressive 1-bit quantization, it needs around 250 GB of combined RAM and VRAM to run at a usable speed, which rules out almost every desktop, laptop, and single-GPU workstation. Ollama's solution is the `:cloud` model tag. Instead of downloading weights, `ollama run kimi-k2.6:cloud` sends your prompt to Moonshot's infrastructure through Ollama's servers and streams the response back to your terminal, using the same commands and API as a local model.

Moonshot has shipped several Kimi K2 versions since the original release. Here's what's live on Ollama as of June 2026:

For most people searching for "Kimi K2 Ollama" today, `kimi-k2.6:cloud` is the tag to use. It's the actively maintained version, has the longest support runway, and adds image input on top of everything the original K2 could already do. Moonshot also offers a hosted Kimi chat assistant if you want to try K2.6 through a web interface without touching a terminal.

Set Up Ollama Cloud and Run Kimi K2.6

Running Kimi K2 through Ollama takes three steps: install Ollama, sign in, and run the model. Nothing here downloads a multi-hundred-gigabyte file. The whole setup takes under five minutes on any machine with a working internet connection.

Step 1: Install Ollama

# Linux and macOS, one-command installer
curl -fsSL https://ollama.com/install.sh | sh

On Windows, download the installer from ollama.com/download, or use winget:

winget install Ollama.Ollama

Verify the installation:

ollama --version
# Expected: ollama version 0.6.x or higher

Cloud models require Ollama 0.6.x or later. If `ollama --version` returns an older release, re-run the install command to update.

Step 2: Sign In to Ollama Cloud

ollama signin

This prints a sign-in URL and opens your browser. Create a free account at ollama.com (or log in if you already have one), then approve the device. The terminal confirms with a message similar to:

Signing in to ollama.com...
Signed in as your-username

Step 3: Run Kimi K2.6 from the Terminal

ollama run kimi-k2.6:cloud

Ollama fetches a small manifest (a few KB, not the model weights, since inference happens remotely), then drops you into a prompt:

pulling manifest
pulling 4f3b2a1c... 100% ▕████████████████▏  3.1 KB
success
>>> Send a message (/? for help)

Type a prompt to test it:

>>> Write a Python function that returns the nth Fibonacci number using memoization.

The first response after signing in can take 10-30 seconds while Ollama establishes the cloud session. After that, responses stream back at normal speed.

Step 4: Verify the Model and Migrate from Older Tags

ollama list

`kimi-k2.6:cloud` appears in the list with a size of a few KB rather than hundreds of gigabytes. That's expected: this is a cloud passthrough entry, not a downloaded model.

Switching Between Local and Cloud Models

`ollama run` works the same way for local and cloud models, so you can keep both on one machine. Pull a small local model alongside Kimi K2.6:

ollama pull qwen3:8b

`ollama list` now shows both `qwen3:8b` (a multi-gigabyte local download) and `kimi-k2.6:cloud` (a manifest-only cloud entry). Switch between them by changing the model name in `ollama run` or in your application's API request. This is useful for keeping a fast local model for routine tasks and reserving Kimi K2.6's larger context and agentic capabilities for harder jobs.

Use Kimi K2 in Your Own Scripts and Agents (API Access)

Beyond the interactive `ollama run` session, Kimi K2.6 is reachable through Ollama's REST API. Any tool that already talks to a local Ollama instance, or to the OpenAI API format, can use it with a one-line model name change.

Generate an API Key

Visit ollama.com/settings/keys while signed in, click "Create API key", and copy the value. Set it as an environment variable:

export OLLAMA_API_KEY=your_api_key_here

Call Kimi K2.6 from the Local Endpoint

curl http://localhost:11434/api/chat -d '{
  "model": "kimi-k2.6:cloud",
  "messages": [
    { "role": "user", "content": "Summarize the difference between Kimi K2 and Kimi K2.6 in two sentences." }
  ],
  "stream": false
}'

Expected output (truncated):

{
  "model": "kimi-k2.6:cloud",
  "message": {
    "role": "assistant",
    "content": "Kimi K2.6 is Moonshot's 1.04T-parameter successor to Kimi K2, adding multimodal image input and swarm-style multi-agent orchestration on top of K2's original 256K-context agentic coding capabilities."
  },
  "done": true
}

Call Kimi K2.6 Directly from ollama.com

For a server or serverless function without a local Ollama install, send requests straight to ollama.com using your API key:

curl https://ollama.com/api/chat \
  -H "Authorization: Bearer $OLLAMA_API_KEY" \
  -d '{
    "model": "kimi-k2.6:cloud",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

Python Example

from ollama import Client

client = Client(host="https://ollama.com", headers={"Authorization": "Bearer " + api_key})

response = client.chat(
    model="kimi-k2.6:cloud",
    messages=[{"role": "user", "content": "Outline a plan to refactor a Flask app into FastAPI."}],
)
print(response["message"]["content"])

OpenAI-Compatible Endpoint for Existing Agent Tools

Ollama exposes an OpenAI-compatible layer at `http://localhost:11434/v1`, the same endpoint used in the Hermes Agent and OpenClaw setups. Point that configuration at `kimi-k2.6:cloud` instead of a local model name, and the agent runs on Kimi K2's 256K context and swarm orchestration without any other config changes:

model:
  default: kimi-k2.6:cloud
  provider: custom
  base_url: http://localhost:11434/v1
  context_length: 256000

Troubleshooting

Frequently Asked Questions