SpaceXAI Releases Grok 4.7: Officially Claims Strongest in Coding/Knowledge Work to Date, Same Price Same Speed Relative to 4.6 Upgrade

SpaceXAI (xAI) officially released Grok 4.7 on September 21, 2026, with the official announcement calling it the most powerful model to date for coding and knowledge work. The company stated that Grok 4.7 offers significant improvements over Grok 4.6 at the same price and speed, and it is available immediately through Cursor, Grok Build, and Grok API. Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed. — SpaceXAI (@SpaceXAI, 2026-09-21) The official page lists pricing starting at US$2 per million input tokens and US$6 per million output tokens. There is also a fast variant, which claims to offer about double the output speed at double the price. Availability spans Cursor, Grok Build, Grok API, third-party coding harnesses, and model routers/cloud platforms. The model card also mentions that the SpaceXAI API can be called via console.x.ai, and that Grok Build defaults to Grok 4.7, with all Cursor plans available. Grok 4.7 works longer on difficult tasks, checks its work more carefully, and comes with our strongest safeguards to date. — SpaceXAI (@SpaceXAI, 2026-09-21) The company described improvements including a larger base model, longer reinforcement learning (RL) training with a focus on tasks requiring multiple hours to complete, stronger self-validation and long context management, as well as native understanding of Grok Bot harnesses. The official comparison table shows that on CursorBench 4.0, Grok 4.7 achieved a score of 46.3%, compared to 40.4% for Grok 4.6 (company-reported figures). On the same day, the GitHub Changelog also announced that Copilot now begins offering Grok 4.7, applicable to Copilot Pro, Pro+, Max, Business, and Enterprise plans, with a gradual rollout. Grok 4.7 is a strong combination of intelligence, speed & low cost — Elon Musk (@elonmusk, 2026-09-21)