DeepSeek is a family of open-weight large language models developed by DeepSeek AI. The R1 series specialises in chain-of-thought reasoning and outperforms many closed models on benchmarks like MATH and AIME. The Coder series is fine-tuned for code generation, completion, and debugging.

Running DeepSeek locally via Ollama means your data never leaves your machine, there are no API costs, and the model works completely offline once downloaded.

macOS version: 11 (Big Sur) or later. Apple Silicon Macs run models natively via Metal GPU acceleration.

After running the install script, open Terminal and start a chat session:

ollama run deepseek-r1:7b

Type your message and press Return. To quit, type /bye or press Ctrl+D.

To list all downloaded models:

ollama list

To remove a model and free disk space:

ollama rm deepseek-r1:7b

Install the Ollama Python library:

pip install ollama

Basic chat example:

import ollama

response = ollama.chat(
    model="deepseek-r1:7b",
    messages=[
        {"role": "user", "content": "Explain quantum entanglement simply."}
    ]
)
print(response["message"]["content"])

Streaming responses:

import ollama

stream = ollama.chat(
    model="deepseek-r1:7b",
    messages=[{"role": "user", "content": "Write a haiku about rain."}],
    stream=True,
)
for chunk in stream:
    print(chunk["message"]["content"], end="", flush=True)

Multi-turn conversation:

import ollama

history = []

def chat(user_msg):
    history.append({"role": "user", "content": user_msg})
    resp = ollama.chat(model="deepseek-r1:7b", messages=history)
    reply = resp["message"]["content"]
    history.append({"role": "assistant", "content": reply})
    return reply

print(chat("What is the capital of Japan?"))
print(chat("What language do they speak there?"))

Ollama exposes an OpenAI-compatible API on http://localhost:11434. Any tool that supports OpenAI can point to it.

Chat completion with curl:

curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-r1:7b",
    "messages": [
      {"role": "user", "content": "Hello, how are you?"}
    ]
  }'

With the official OpenAI Python SDK:

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama",  # required but ignored
)

response = client.chat.completions.create(
    model="deepseek-r1:7b",
    messages=[{"role": "user", "content": "Summarise the Apollo 11 mission."}],
)
print(response.choices[0].message.content)

Install the Ollama JS library:

npm install ollama

Basic usage:

import ollama from "ollama";

const response = await ollama.chat({
  model: "deepseek-r1:7b",
  messages: [{ role: "user", content: "What is 17 × 34?" }],
});

console.log(response.message.content);

Streaming in Node.js:

import ollama from "ollama";

const stream = await ollama.chat({
  model: "deepseek-r1:7b",
  messages: [{ role: "user", content: "Count to 10 slowly." }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.message.content);
}

Create a Modelfile to give your model a persistent system prompt or tweak parameters:

FROM deepseek-r1:7b

SYSTEM """
You are a senior software engineer. Always respond with clean,
commented code. Prefer Python unless asked otherwise.
"""

PARAMETER temperature 0.2
PARAMETER top_p 0.9

Build and run your custom model:

ollama create my-coder -f ./Modelfile
ollama run my-coder

Useful parameters:

Ollama command not found after install

Close Terminal and reopen it, or run source ~/.zshrc. The installer adds Ollama to /usr/local/bin.

Model pull fails or stalls

Check your internet connection. Resume an interrupted pull by running ollama pull MODEL_ID again — it resumes from where it stopped.

"Error: model requires more system memory"

Your Mac does not have enough RAM. Choose a smaller model (e.g. 1.5B or 7B) or close other applications to free memory.

Responses are very slow

On Intel Macs without a GPU, inference is CPU-only. Try a smaller model. Apple Silicon Macs use Metal GPU automatically for much faster speeds.

Port 11434 already in use

Another process is using the Ollama port. Run lsof -i :11434 to find it, or set a custom port: OLLAMA_HOST=0.0.0.0:11435 ollama serve.

How do I update Ollama?

Re-run the install script or download the latest version from ollama.com. Your downloaded models are preserved.