Grok 4.7 lands in Grok Build: 4 reasoning tiers, Terminal-Bench jumps 20% to 38%

XFreeze · x · 2026-09-22

Grok 4.7 is live in Grok Build with selectable Low/Medium/High/Extra High reasoning effort. Official benchmarks vs Grok 4.6: CursorBench 4.0 40.4%→46.3%, Terminal-Bench 4.0 20.3%→38.0%, EEBench 53.0%→64.0%, HealthBench Professional 48.5%→56.7%. Same price and speed, with a larger base model, longer RL training, better self-verification, and much stronger long-horizon reasoning.

Related event: xAI Launches Grok 4.7, Its Strongest Coding Model Yet(26 posts)→

Original post →

More from Models

Models channel →