OpenAI API Compatibility

Drop-in replacement for OpenAI SDK. Use existing OpenAI code with Ollama models.

Setup

Base URL: http://localhost:11434/v1/
API Key:  "ollama"  (required by SDK but ignored by server)

Models must be pulled locally first: ollama pull qwen3:1.7b

Supported Endpoints

EndpointFeatures POST /v1/chat/completionsStreaming, JSON mode, vision, tools, thinking POST /v1/completionsStreaming, JSON mode POST /v1/embeddingsString/array input, dimensions POST /v1/images/generationsExperimental, b64_json only POST /v1/responsesTools, reasoning (non-stateful) GET /v1/modelsList available models

Not Supported

logprobs in chat completions

tool_choice (model decides)

Stateful responses

Python (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url='http://localhost:11434/v1/',
    api_key='ollama',
)
Chat completion
response = client.chat.completions.create(
    model='qwen3:1.7b',
    messages=[{'role': 'user', 'content': 'Why is the sky blue?'}],
)
print(response.choices[0].message.content)

Streaming
stream = client.chat.completions.create(
    model='qwen3:1.7b',
    messages=[{'role': 'user', 'content': 'Tell me a story'}],
    stream=True,
)
for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end='')

Vision
response = client.chat.completions.create(
    model='gemma3',
    messages=[{
        'role': 'user',
        'content': [
            {'type': 'text', 'text': "What's in this image?"},
            {'type': 'image_url', 'image_url': 'data:image/png;base64,...'},
        ],
    }],
    max_tokens=300,
)

Embeddings
response = client.embeddings.create(
    model='mxbai-embed-large',
    input=['Hello world', 'Goodbye world'],
)
print(len(response.data[0].embedding))

Thinking (reasoning_effort)
response = client.chat.completions.create(
    model='deepseek-r1:1.5b',
    messages=[{'role': 'user', 'content': 'Solve: 17 * 23'}],
    reasoning_effort='high',  # "high", "medium", "low", "none"
)

JavaScript (OpenAI SDK)

import OpenAI from 'openai'

const client = new OpenAI({
    baseURL: 'http://localhost:11434/v1/',
    apiKey: 'ollama',
})
const response = await client.chat.completions.create({
    model: 'qwen3:1.7b',
    messages: [{ role: 'user', content: 'Why is the sky blue?' }],
})
console.log(response.choices[0].message.content)

cURL

curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3:1.7b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Cloud via OpenAI SDK

from openai import OpenAI

client = OpenAI(
    base_url='https://ollama.com/v1/',
    api_key=os.environ['OLLAMA_API_KEY'],
)
response = client.chat.completions.create(
    model='gpt-oss:120b',
    messages=[{'role': 'user', 'content': 'Explain quantum computing'}],
)

Model Aliases for Compatibility

Some tools expect OpenAI model names. Create aliases:

ollama cp qwen3:1.7b gpt-3.5-turbo
ollama cp qwen3.5:27b gpt-4

mindX Integration

mindX already uses multiple LLM providers via llm_factory.py. The OpenAI-compatible endpoint means any provider handler written for OpenAI works with Ollama:

# In llm_factory.py, Ollama can serve as a drop-in for OpenAI
by pointing base_url to localhost:11434/v1/
This enables using the same OpenAI handler code for local inference

---