Claude Code (local backend) with a local model: setup for Ollama or LM Studio

Claude Code is Anthropic's proprietary terminal coding agent; pointed at a local server's Anthropic-compatible API, it runs against open-weight models. Ollama and LM Studio both document this setup, and it suits people already using Claude Code who have the memory for a 64k+ context.

Source code Official docs Compare all agents

At a glance

Type: cli
Licence: Proprietary
Latest release: 2.1.303 (2026-10-18)
Status: Maintained
Local backends: ollama, lmstudio, anthropic-compatible
Needs native tool calling: Yes
MCP support: Yes
Recommended minimum context: 64,000 tokens
Start-up prompt: ~33,000 tokens (single third-party measurement)
GitHub stars: Not verified

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 (macOS / Linux):

curl -fsSL https://claude.ai/install.sh | bash

Start Ollama with a 64k context window (or set the context slider in the Ollama app's settings).
```
OLLAMA_CONTEXT_LENGTH=64000 ollama serve
```
In a second terminal, pull a tool-capable coding model (replace qwen3-coder:30b with any model you prefer).
```
ollama pull qwen3-coder:30b
```
Quickest path: let Ollama configure and launch Claude Code.
```
ollama launch claude
```

Or point Claude Code at Ollama's Anthropic-compatible API yourself.

export ANTHROPIC_AUTH_TOKEN=ollama
export ANTHROPIC_API_KEY=""
export ANTHROPIC_BASE_URL=http://localhost:11434

Tell Claude Code the real context window, since it assumes 200K for model IDs it doesn't recognize.
```
export CLAUDE_CODE_MAX_CONTEXT_TOKENS=64000
```
Run Claude Code with your local model.
```
claude --model qwen3-coder:30b
```

Other ways to install

Windows (PowerShell)

irm https://claude.ai/install.ps1 | iex

Check that it works

Inside Claude Code, run /status to check which base URL and credentials are active.

/status

Which local models fit, by machine

Checked by this site's engine at the context Claude 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	64,000	Comfortable (8.4 GB of 12 GB)	`qwen3.5:9b`
	Qwen 3.6 35B-A3B	Q4_K_M	64,000	With CPU offload (MoE) (23.3 GB of 12 GB)	`qwen3.6:35b-a3b`
16 GB GPU	Qwen 3.5 9B	Q4_K_M	64,000	Comfortable (8.4 GB of 16 GB)	`qwen3.5:9b`
	Qwen 3.6 35B-A3B	Q4_K_M	64,000	With CPU offload (MoE) (23.3 GB of 16 GB)	`qwen3.6:35b-a3b`
24 GB GPU	Qwen3.8 27B	Q4_K_M	64,000	Tight (21.9 GB of 24 GB)	`qwen3.8:27b`
	Qwen 3.6 27B	Q4_K_M	64,000	Tight (21.9 GB of 24 GB)	`qwen3.6:27b`
	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	64,000	Comfortable (21.9 GB of 32 GB)	`qwen3.8:27b`
	Qwen 3.6 27B	Q4_K_M	64,000	Comfortable (21.9 GB of 32 GB)	`qwen3.6:27b`
	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`
	Devstral Small 2 24B	Q4_K_M	64,000	Comfortable (25.8 GB of 32 GB)	`devstral-small-2:24b`
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

Ollama recommends a 64k+ context window for Claude Code; LM Studio says more than ~25k. Claude Code's system prompt alone is large, so stay at 64k where you can.
Ollama's Anthropic-compatible API does not support tool_choice or prompt caching (cache_control).
Claude settings files can override shell variables; if the connection goes elsewhere, update the base URL or credentials there and re-check with /status.
If LM Studio's Require Authentication is on, set ANTHROPIC_AUTH_TOKEN to your LM Studio API token instead of lmstudio.

Other coding agents

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

Sources (7), checked 2026-10-20