Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/dais-polymtl/flock/llms.txt

Use this file to discover all available pages before exploring further.

Ollama lets you run open-weight models locally without any external API calls or API keys. Flock connects to your local Ollama instance over HTTP, so queries stay entirely on your machine. This makes Ollama a good choice for development, offline use, or privacy-sensitive workloads. Both text completion and vision models are supported.

​
Prerequisites

Before configuring Flock, you need Ollama installed and running with at least one model downloaded:
  1. Install Ollama — download from ollama.com/download
  2. Pull a model — for example: ollama pull llama3.2
  3. Confirm Ollama is running — the default address is 127.0.0.1:11434
Make sure Flock is installed and loaded before continuing — see the Quickstart if you haven’t done that yet.

​
Configure your secret

Ollama does not require an API key. You only need to tell Flock where Ollama is listening: 
CREATE SECRET (
    TYPE ollama,
    API_URL '127.0.0.1:11434'
);
The API_URL field is required and must point to your running Ollama instance. If you’re running Ollama on a different host or port, update the URL accordingly.
If you change the Ollama port with the OLLAMA_HOST environment variable, update API_URL here to match.

​
Create a model

Register a named model in Flock using the exact model name you pulled with Ollama: 
CREATE MODEL(
    'QuackingModel',
    'llama3.2',
    'ollama',
    {"tuple_format": "json", "batch_size": 32, "model_parameters": {"temperature": 0.7}}
);
The four arguments are:
ArgumentDescription
'QuackingModel'Unique name you reference in queries
'llama3.2'Ollama model name (must be already pulled)
'ollama'Provider name
{...}Config: batch size, tuple format, and model parameters
The model name must exactly match what you pulled with ollama pull. Run ollama list to see all downloaded models on your system.

​
Run a query

With your secret and model in place, call llm_complete: 
SELECT llm_complete(
    {'model_name': 'QuackingModel'},
    {'prompt': 'Write a short poem about a database.'}
);
To use column data as context: 
SELECT llm_complete(
    {'model_name': 'QuackingModel'},
    {
        'prompt': 'Classify the topic of this article: {article}',
        'context_columns': [{'data': article_text, 'name': 'article'}]
    }
) AS topic
FROM articles;

​
Supported model types

Any Ollama text/chat model works with llm_complete, llm_filter, and aggregate functions. Popular choices include:
ModelPull command
Llama 3.2 (3B)ollama pull llama3.2
Llama 3.1 (8B)ollama pull llama3.1
Mistral 7Bollama pull mistral
Gemma 2 (9B)ollama pull gemma2
Phi-3 Miniollama pull phi3
See the full catalog at ollama.com/library.

​
Next steps

Image support

Analyze images with vision models in SQL queries

Scalar functions

Full reference for llm_complete, llm_filter, llm_embedding, and more

Build docs developers (and LLMs) love

Get started for freeTalk to us