SpaceXAI Launches Grok 4.7: Reasoning Model for Coding and Long-Duration Tasks

SpaceXAI has released Grok 4.7, a new base model powering all tiers of its Grok assistant from fast to heavy modes, rolling out across web and mobile apps on September 21. The company built the model with a longer reinforcement-learning run focused on demanding multi-hour tasks, doubled down on self-verification capabilities, and optimized it for longer-context work across its 500,000-token window. Pricing holds steady at $2 per million input tokens and $6 per million output tokens—matching Grok 4.6 despite the larger base model.