Grok 3 mini

Grok 3 Mini is a compact Grok-3 variant optimized for speed and utility on logic-oriented tasks. It explicitly “thinks before responding” and lets you view raw reasoning traces, making it useful for audits and teaching agents. The model supports function calling (tool use) and structured outputs for reliable integrations, and provides a 131,072-token context window for longer prompts and documents. Pricing is $0.30 per million input tokens (with $0.075/M for cached input) and $0.50 per million output tokens; optional Live Search is billed per 1K sources. In addition to the standard endpoint, a grok-3-mini-fast variant is offered; Oracle’s catalog notes both use the same underlying model with identical response quality, differing only in serving. Microsoft’s model catalog further highlights adjustable “low” and “high” thinking budgets to trade off depth versus latency. Grok 3 Mini is available via the xAI API and through major cloud catalogs (e.g., Azure AI and Oracle OCI).