Grok 4.3

Grok 4.3 is a reasoning model in SpaceXAI's Grok family, built to be fast and dependable. It follows instructions closely and calls tools reliably, which keeps multi-step requests on track. It reads text and images and holds up to 1,000,000 tokens in context, so you can paste a long log, a full spec or a stack of screenshots into one conversation.

Teams use it for steady day-to-day technical work: explaining an unfamiliar module, writing and fixing functions, turning a spreadsheet export into a summary, reading a chart or UI screenshot, or answering questions over a long contract or report. Close instruction following also makes it a good fit when the output has to match a set format, such as JSON or a fixed template.

Grok 4.7 is the newer flagship in the family, trained for long coding and agent tasks that run for hours, with a 500,000-token window. Grok 4.3 keeps the larger 1M-token window and quicker turns. Both run in Endless chat, so you can move between them in the same workspace as a task grows.