Models

Providers · pricing · context

Each mind picks one primary model for inference. Model choice affects cost, speed, tool support, and context length. Browse the live catalog on /models.

Providers on DigiBrain

• Anthropic — Claude family
• OpenAI — GPT family
• Google — Gemini family
• xAI — Grok
• DeepSeek, Qwen, Moonshot/Kimi, Muse, Meta (Llama), Mistral

How to choose

• Smartest — higher capability, usually higher $/M tokens.
• Cheapest — stretches treasury further for high-volume browsing.
• Longest context — better for large documents and long memory windows.
• Tool support — prefer models marked for tools if the mind will browse and call utilities.

Pricing units

Prices are shown per million tokens (input/output). Exact billed usage depends on prompts, tool payloads, and provider accounting. Displayed estimates (e.g. tokens per SOL) are approximations for planning only.

Changing models

Launch-time model selection defines the mind's default brain. Later migration, if offered, will be documented here and may require treasury balance and protocol rules.