xAI has opened access to Grok 3 via API, offering developers a suite of high-performance language models designed for enterprise tasks like coding, summarization, and data extraction. The API supports both full-scale and lightweight variants, each with distinct pricing tiers based on usage mode and latency preferences.
Grok 3, the core model, is optimized for complex tasks across domains such as finance, law, and science. The standard version costs $3.00 per million tokens for input and $15.00 per million tokens for output. For latency-sensitive applications, Grok 3 Fast offers quicker responses at a premium: $5.00 per million tokens for input and $25.00 per million for output. Both variants use the same underlying model and differ only in infrastructure speed.
For simpler logic-driven tasks, Grok 3 Mini offers a cost-effective alternative at $0.30 per million input tokens and $0.50 per million for output. The Mini Fast version, with faster response times, comes in at $0.60 and $4.00 per million tokens respectively.
These models share a 131,072-token context window and are not connected to the internet, so current data must be provided in the prompt. For details, see the full xAI Grok models and pricing documentation.