Kilo Code with a local model: setup for Ollama or LM Studio

Kilo Code is an open-source (MIT) coding agent for VS Code, JetBrains and the terminal; the Kilo CLI is a fork of OpenCode. It has Ollama and LM Studio providers configured in the settings UI or kilo.json, and its docs suggest a 24 GB+ GPU or a 32 GB+ Mac for local models.

Source code Official docs Compare all agents

At a glance

Type: vscode-extension, jetbrains, cli
Licence: MIT
Latest release: 7.8.14 (extension); 7.8.12 (CLI) (2026-10-16)
Status: Maintained
Local backends: ollama, lmstudio
Needs native tool calling: Yes
MCP support: Yes
Recommended minimum context: 32,000 tokens
Start-up prompt: Not measured
GitHub stars: 27.6k (2026-10-19)

How to set it up with a local model

Pick your backend. Each command has its own Copy button. Where a step uses an example model, swap in the one you run; the table below lists the ones that fit your machine.

Install it in VS Code: Extensions view → search "Kilo Code" → Install.
Start Ollama with a 32k context window (or set the context slider in the Ollama app's settings).
```
OLLAMA_CONTEXT_LENGTH=32768 ollama serve
```
In a second terminal, pull Kilo's recommended model (replace it with any tool-capable model you prefer).
```
ollama pull qwen3-coder:30b
```
In the extension, open Settings → Providers and add Ollama (no API key needed).
Set "Context Window Size (num_ctx)" to at least 32k in the provider settings.

For the CLI (or a model missing from the picker), register the model in kilo.json with tool calling and its context limit.

~/.config/kilo/kilo.json

{
  "model": "ollama/qwen3-coder:30b",
  "provider": {
    "ollama": {
      "models": {
        "qwen3-coder:30b": {
          "name": "Qwen3-Coder 30B",
          "tool_call": true,
          "limit": {
            "context": 32768,
            "output": 8192
          }
        }
      }
    }
  }
}

Other ways to install

JetBrains: Settings → Plugins → search "Kilo Code" → Install

CLI (npm)

npm install -g @kilocode/cli

CLI (macOS / Linux)

curl -fsSL https://kilo.ai/cli/install | bash

Check that it works

Select the ollama/ or lmstudio/ model in Kilo and ask it to list the files in the workspace. If you see an ollama-cloud/... model id, you are on the hosted Kilo Gateway provider rather than your local server.

ollama ps

Which local models fit, by machine

Checked by this site's engine at the context Kilo Code needs, with 32 GB of system RAM for MoE offload. A Mac counts three quarters of unified memory as usable.

Machine	Model	Quant	Context	Fit	Ollama
12 GB GPU	Qwen 3.5 9B	Q4_K_M	32,768	Comfortable (7.4 GB of 12 GB)	`qwen3.5:9b`
	Qwen 3.6 35B-A3B	Q4_K_M	32,768	With CPU offload (MoE) (22.7 GB of 12 GB)	`qwen3.6:35b-a3b`
16 GB GPU	Qwen 3.5 9B	Q4_K_M	32,768	Comfortable (7.4 GB of 16 GB)	`qwen3.5:9b`
	Qwen 3.6 35B-A3B	Q4_K_M	32,768	With CPU offload (MoE) (22.7 GB of 16 GB)	`qwen3.6:35b-a3b`
24 GB GPU	Qwen3.8 27B	Q4_K_M	32,768	Comfortable (19.9 GB of 24 GB)	`qwen3.8:27b`
	Qwen 3.6 27B	Q4_K_M	32,768	Comfortable (19.9 GB of 24 GB)	`qwen3.6:27b`
	Devstral Small 2 24B	Q4_K_M	32,768	Comfortable (20.7 GB of 24 GB)	`devstral-small-2:24b`
	Qwen 3.6 35B-A3B	Q4_K_M	65,536	Tight (23.3 GB of 24 GB)	`qwen3.6:35b-a3b`
32 GB GPU	Qwen3.8 27B	Q4_K_M	32,768	Comfortable (19.9 GB of 32 GB)	`qwen3.8:27b`
	Qwen 3.6 27B	Q4_K_M	32,768	Comfortable (19.9 GB of 32 GB)	`qwen3.6:27b`
	Devstral Small 2 24B	Q4_K_M	32,768	Comfortable (20.7 GB of 32 GB)	`devstral-small-2:24b`
	Qwen 3.6 35B-A3B	Q4_K_M	65,536	Comfortable (23.3 GB of 32 GB)	`qwen3.6:35b-a3b`
	Qwen3-Coder 30B-A3B (MoE)	Q4_K_M	65,536	Comfortable (25.7 GB of 32 GB)	`qwen3-coder:30b`
Mac, 64 GB	Qwen 3.6 35B-A3B	Q8_0	131,072	Comfortable (40.7 GB of 48 GB)	—

Your exact machine: filter the matrix or check any model in the analyzer.

Common problems with local models

Kilo's docs say you need at least 32k context for decent results; Ollama truncates prompts to a short window by default.
Kilo's docs warn qwen3-coder:30b sometimes fails to call tools correctly; try rewording the prompt or the Enhance Prompt button, or devstral:24b as an alternative.
Keeping conversations short and disabling MCP tools can speed up local models.
Requests time out after 10 minutes by default; raise API Request Timeout in the extension settings for slow local models.

Other coding agents

How they rank for local models: the best agent harness for local LLMs.

Sources (5), checked 2026-10-20