AI Engine
StreamBuddy supports two AI providers. Pick based on your hardware and budget.
Local — Ollama
Best for | Privacy, no API cost, offline use |
|---|---|
Requires | Ollama (auto-installed by the StreamBuddy installer) |
Recommended model | llama3.2:3b — fast, good quality, low RAM |
Default URL | http://localhost:11434 |
Note: The first response after launch is slow (Model cold-start). Wait for the green AI status indicator on the Dashboard before going live, replies after that are fast.
Tip: Keep Max Response Length around 80-120 tokens. Shorter replies feel snappy in chat; longer ones read like essays out loud.
Cloud — Claude API
Best for | Higher-quality responses, no GPU required |
|---|---|
Requires | An Anthropic API key from console.anthropic.com |
Recommended model | Claude Haiku 4.5 — fastest and cheapest for live chat |
Tip: Set Response Rate to 20-40% on busy channels to control API spend. At 100% on an active stream, credits disappear fast.
Hardware Auto-Detect
Settings can scan your GPU and RAM and suggest the Ollama model that best fits your hardware, then download it for you.