LLM Setup

GenDB uses an LLM to analyze your database schema and produce a generation plan — a mapping of each column to an appropriate data generator. This page covers how to configure different LLM providers.

How GenDB Uses LLMs

The LLM generates all data values directly. GenDB sends your schema (table names, column names, data types, constraints, foreign key relationships) to the LLM, which returns realistic, semantically coherent data as JSON. Data is generated in batches per table (up to 50 rows per LLM call), with larger tables chunked automatically.

Ollama (Local)

Ollama is the default provider and runs entirely on your machine.

Setup

1. Install Ollama: ollama.com

2. Pull a model:

   ollama pull qwen2.5:7b
   

3. GenDB uses Ollama by default — no configuration changes needed:

   llm:
     provider: ollama
     model: qwen2.5:7b
     base_url: http://localhost:11434/v1
   

Tip

Qwen 2.5 7B is the recommended default — it produces reliable structured JSON output and runs well on machines with 8GB+ RAM. For constrained hardware (4GB RAM), use llama3.2 (3B parameters) instead, but expect lower JSON reliability.

OpenAI

Setup

1. Get an API key from platform.openai.com

2. Configure in gendb.yaml:

   llm:
     provider: openai
     model: gpt-4o-mini
     api_key: sk-...
   

Tip

gpt-4o-mini is recommended for a good balance of quality and cost. Schema analysis is a single API call per generate_data, so costs are minimal.

Custom / Self-Hosted

Any OpenAI-compatible API endpoint works — including vLLM, text-generation-inference, DeepSeek, and other providers.

llm:
  provider: custom
  model: deepseek-coder
  base_url: http://localhost:8000/v1
  api_key: ""  # if required by your endpoint

Column Instructions

You can provide per-column instructions in gendb.yaml to guide the LLM:

generation:
  tables:
    users:
      columns:
        bio:
          prompt: "Write a short professional bio"
        status:
          generator: one_of
          values: [active, inactive, pending]

These instructions are included in the LLM prompt alongside the schema context.

Cost Profile

Operation	LLM calls	Typical cost (OpenAI)
Data generation	1 per 50 rows per table	~$0.01-0.05 per table

With Ollama, all operations are free and local.