Open vs Closed Models

Closed models (GPT-4, Claude, Gemini) are accessed via API: easy to use, always updated, but you can't inspect or self-host them, and your data leaves your infrastructure.

Open-weight models (Llama, Mistral, Qwen, DeepSeek) publish their weights: you can download, fine-tune, and run them anywhere — including fully offline. 'Open' varies: weights may be open while training data and code stay closed.

The trade-off: closed models usually lead on raw capability and convenience; open models win on cost at scale, privacy, customization, and independence from any single vendor.