Grok 4.7

Grok 4.7 is a conversational large language model in the Grok series, credited in our database to SpaceXAI (xAI). It sits in the Grok 4.x generation of the family, a point release that follows the Grok 4 line rather than a separate lightweight or reasoning-only variant. Our catalog classifies it as a chat model, meaning it is intended for instruction-following, dialogue, and general assistant workloads exposed over an inference API. The measurements we track for Grok 4.7 come from Artificial Analysis: roughly 38.35 output tokens per second and a time to first token of about 712 ms. Those figures describe throughput and latency as observed on a hosted endpoint, not model architecture, and they can shift with provider, region, load, and prompt length. We do not currently track a confirmed context window, modality list, or public benchmark scores for this specific release, so any such details should be checked against the serving provider's own model card before you commit to it in production. In practice, models in this tier are used for general chat assistants, drafting and summarization, coding help, and agent-style loops where a single capable general-purpose model handles most steps. When comparing Grok 4.7 against peers, the useful axes are the measured latency and throughput above, the per-provider rates in the pricing table on this page, and your own evaluation on representative prompts — published family-level claims rarely transfer cleanly to a specific point release.