SpaceXAI Introduces Grok 4.7 Model with New Pricing and Features

The SpaceXAI release notes now feature the Grok 4.7 model, superseding Grok 4.6 as the latest offering for developers. This update, dated September 21, 2026, makes Grok 4.7 available through the xAI API under the identifier `grok-4.7`. The model is designed for coding, agentic tasks, and knowledge work, offering a substantial 500,000 token context window. It accepts both text and image inputs, producing text-only output without a specified text output limit.

Pricing for Grok 4.7 has been adjusted: for prompts under 200,000 tokens, input tokens are $2 per million, cached input tokens are $0.50 per million, and output tokens are $6 per million. For prompts exceeding 200,000 tokens, these rates increase to $4, $1, and $12 per million tokens, respectively. The model supports varying reasoning efforts: low, medium, high (default), and xhigh. When using the Responses API, `reasoning.encrypted_content` is always returned, even if not explicitly requested. Grok 4.7 is accessible via the US regional endpoint. A faster variant, Grok 4.7 Fast, is mentioned but is exclusively available through Cursor and Grok Build, not the public xAI API.