How to Integrate Claude Code with Ollama for Local and Cloud LLM Workflows

Overview

Claude Code can now use Ollama as a backend, enabling local or cloud LLMs for private, cost‑free development.

Installation

Claude Code

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

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

curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd   # CMD

Verify with claude --version .

Ollama

curl -fsSL https://ollama.com/install.sh | sh   # macOS/Linux/WSL

Windows users download the installer from https://ollama.com/download and run it. Verify with ollama --version . Ollama runs as a service at http://localhost:11434 .

Pull a model ollama pull qwen3-coder Other useful models: ollama pull qwen2.5-coder:7b – balanced 5 GB coding model ollama pull starcoder2:3b – compact 1.7 GB model ollama pull qwen2.5-coder:1.5b – lightweight 1 GB model ollama pull deepseek-coder:1.3b – smallest 776 MB (no tool calls)

Confirm with ollama list .

Configure Environment Variables

Point Claude Code to the local Ollama endpoint:

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

PowerShell equivalent:

$env:ANTHROPIC_AUTH_TOKEN="ollama"
$env:ANTHROPIC_BASE_URL="http://localhost:11434"

Persist these in .bashrc or .zshrc for future sessions.

Run Claude Code with a Model

Start Claude Code specifying the model name: claude --model qwen2.5-coder:7b If a model (e.g., deepseek-coder:1.3b) does not support tool calls, switch to a model that does, such as qwen2.5-coder:7b.

Context Length Considerations

Claude Code works best with models offering at least a 32k token window; 64k is ideal for large projects. Adjust the context via a custom Modelfile:

FROM qwen2.5-coder:7b
PARAMETER num_ctx 32768

Create the custom model and use it:

ollama create qwen-32k -f Modelfile
claude --model qwen-32k

Typical context sizes:

2048 tokens – default for many small models

8192 tokens – simple tasks

32768 tokens – Claude Code minimum

65536 tokens – complex projects

Hardware Requirements & Performance

~16 GB RAM comfortably runs 7 b parameter models.

20 b+ models benefit from 32 GB + RAM.

Apple Silicon provides strong performance; older hardware may struggle with larger models.

7 b models are fast enough for day‑to‑day coding; 20 b models are slower but handle broader tasks such as documentation or architecture planning.

Using Ollama Cloud (Hosted Models)

When local resources are insufficient, register at https://ollama.com/cloud and obtain an API key. Set the variables:

export ANTHROPIC_BASE_URL=https://ollama.com
export ANTHROPIC_API_KEY=your-api-key-here

PowerShell:

$env:ANTHROPIC_BASE_URL="https://ollama.com"
$env:ANTHROPIC_API_KEY="your-api-key-here"

Run a cloud model by appending :cloud to the name, e.g.: claude --model glm-4.7:cloud Cloud models run with full context length and require no local storage.

Tool‑Calling Example

The following Python snippet shows a tool call (weather lookup) using the local Ollama endpoint:

import anthropic
client = anthropic.Anthropic(
    base_url='http://localhost:11434',
    api_key='ollama',
)
message = client.messages.create(
    model='qwen3-coder',
    max_tokens=1024,
    tools=[{
        'name': 'get_weather',
        'description': 'Get the current weather in a location',
        'input_schema': {
            'type': 'object',
            'properties': {'location': {'type': 'string', 'description': 'The city and state, e.g. San Francisco, CA'}},
            'required': ['location']
        }
    }],
    messages=[{'role': 'user', 'content': "What's the weather in San Francisco?"}]
)
for block in message.content:
    if block.type == 'tool_use':
        print(f'Tool: {block.name}')
        print(f'Input: {block.input}')

Key Takeaways

Integrating Claude Code with Ollama provides a flexible, privacy‑preserving workflow. Local models are ideal for private codebases and rapid iteration, while Ollama Cloud offers scalable performance for larger contexts or hardware‑limited setups.