Model Parameters

Flock allows you to configure model behavior through the model_parameters field in LLM function calls. This provides fine-grained control over how models generate responses, enabling you to optimize performance for specific use cases. This page is part of Resource Management. For batching and quotas, see Models.

Overview

Model parameters are passed within the model_parameters field—either in CREATE MODEL / UPDATE MODEL statements or inline in LLM function calls. Different providers support different parameters, allowing you to customize temperature, token limits, sampling methods, and more. For model configuration options such as max_batch_size, is_async, and rate_limit, see Models Management. Compatibility: Works with all Flock LLM functions - llm_complete, llm_filter, llm_embedding, llm_reduce, llm_rerank, llm_first, llm_last Each provider expresses structured JSON differently in model_parameters (for example OpenAI response_format, Ollama format, Anthropic output_format / tool-use). Check the Output section on each function page for examples — start with llm_complete and llm_reduce — and the provider pages under Getting Started for API-specific details.

OpenAI Parameters

OpenAI models support a comprehensive set of parameters for controlling generation behavior. For complete parameter reference, see OpenAI Chat Completions API.

Syntax

{
  'model_name': 'gpt-4o',
  'model_parameters': '{
    "temperature": 0.7,
    "max_tokens": 1000,
    "top_p": 1.0,
    "frequency_penalty": 0.0,
    "presence_penalty": 0.0,
    "stop": ["\\n", "END"],
    "response_format": {...}
  }'
}

Example Usage

SELECT llm_complete(
  {
    'model_name': 'gpt-4o',
    'model_parameters': '{
        "temperature": 0.7,
        "max_tokens": 1000,
        "top_p": 1.0,
        "frequency_penalty": 0.0,
        "presence_penalty": 0.0,
        "stop": ["\\n", "END"],
        "response_format": {...}
    }'
  },
  { 'prompt': 'Write a professional email subject line.' },
  { 'topic': 'quarterly report' }
) AS response;

Ollama Parameters

Ollama models support different parameters optimized for local deployment. For complete parameter reference, see Ollama Chat Completions API.

Syntax

{
  'model_name': 'llama3.1',
  'model_parameters': '{
    "temperature": 0.7,
    "num_predict": 1000,
    "top_p": 0.9,
    "top_k": 40,
    "repeat_penalty": 1.1,
    "format": {...}
  }'
}

Example Usage

SELECT llm_complete(
  {
    'model_name': 'llama3.1',
    'model_parameters': '{
      "temperature": 0.3,
      "num_predict": 300
    }'
  },
  { 'prompt': 'Provide a factual explanation.' },
  { 'topic': 'photosynthesis' }
) AS explanation;

Azure OpenAI Parameters

Azure OpenAI supports the same parameters as OpenAI. For complete parameter reference, see OpenAI Chat Completions API.

Syntax

{
  'model_name': 'gpt-4o',
  'model_parameters': '{
    "temperature": 0.7,
    "max_tokens": 1000,
    "top_p": 1.0,
    "frequency_penalty": 0.0,
    "presence_penalty": 0.0,
    "stop": null
  }'
}

Common Parameter Patterns

Deterministic Output

SELECT llm_complete(
  {
    'model_name': 'gpt-4o',
    'model_parameters': '{
      "temperature": 0.0,
      "max_tokens": 500
    }'
  },
  { 'prompt': 'Explain the concept accurately.' },
  { 'concept': 'quantum computing' }
) AS factual_explanation;

Creative Generation

SELECT llm_complete(
  {
    'model_name': 'gpt-4o',
    'model_parameters': '{
      "temperature": 0.9,
      "top_p": 0.95,
      "presence_penalty": 0.5,
      "max_tokens": 800
    }'
  },
  { 'prompt': 'Write a creative story.' },
  { 'genre': 'science fiction', 'setting': 'mars colony' }
) AS creative_story;

Code Generation

SELECT llm_complete(
  {
    'model_name': 'gpt-4o',
    'model_parameters': '{
      "temperature": 0.0,
      "max_tokens": 300,
      "stop": ["\\n\\n", "```"]
    }'
  },
  { 'prompt': 'Generate clean, working code.' },
  { 'language': 'python', 'task': 'sort a list' }
) AS code_snippet;

Anthropic Parameters

Anthropic Claude models use the Anthropic Messages API. Common parameters:

{
  'model_name': 'claude-sonnet-4-5',
  'model_parameters': '{
    "temperature": 0.7,
    "max_tokens": 1024,
    "system": "You are a helpful data analyst assistant.",
    "top_p": 0.9
  }'
}

Home

Getting Started

Resource Management

Scalar Functions

Aggregate Functions

Hybrid Search

Operations

Contributing

Model Parameters

Overview

OpenAI Parameters

Syntax

Example Usage

Ollama Parameters

Syntax

Example Usage

Azure OpenAI Parameters

Syntax

Common Parameter Patterns

Deterministic Output

Creative Generation

Code Generation

Anthropic Parameters

​Overview

​OpenAI Parameters

​Syntax

​Example Usage

​Ollama Parameters

​Syntax

​Example Usage

​Azure OpenAI Parameters

​Syntax

​Common Parameter Patterns

​Deterministic Output

​Creative Generation

​Code Generation

​Anthropic Parameters

Overview

OpenAI Parameters

Syntax

Example Usage

Ollama Parameters

Syntax

Example Usage

Azure OpenAI Parameters

Syntax

Common Parameter Patterns

Deterministic Output

Creative Generation

Code Generation

Anthropic Parameters