xAI

Artificial Intelligence · Developer Tools

Launch

xAI launches Grok 4.7, its most capable coding and knowledge model

Independent tests show it trailing Claude and GPT-6 on agentic coding, despite pricing that undercuts both by a wide margin.

• xAI released Grok 4.7 on September 21, 2026, calling it its most capable model yet for coding and knowledge work, priced at $2 per million input tokens and $6 per million output tokens—unchanged from Grok 4.6.
• Elon Musk's xAI is pitching Grok 4.7 as a direct rival to OpenAI's GPT-6 and Anthropic's Claude Fable 5.1, shipping it into coding tools like Cursor, Grok Build and GitHub Copilot.
• The model runs on a larger base model, reportedly 2.1 trillion parameters (up 40% from Grok 4.6's 1.5 trillion), trained with a longer reinforcement-learning run on multi-hour tasks and SpaceX engineering data.
• xAI's own benchmarks show Grok 4.7 scoring 46.3% on CursorBench 4.0, ahead of Grok 4.6 (40.4%) and GPT-5.6 Sol Max (41.7%), but behind Claude Fable 5.1's 51.8%.
• On the independent Artificial Analysis Intelligence Index, Grok 4.7 scores 46 overall and just 26% on Terminal-Bench 4.0's agentic coding test, versus 60% for GPT-6 Astra and 55% for Claude Fable 5.1.
• Grok 4.7 lost the CursorBench comparison to Claude Fable 5.1 even though Cursor, the benchmark's creator, is the coding editor xAI's parent company had recently acquired, according to Yellow.
• The launch signals xAI is competing on price rather than raw capability, undercutting rivals' token costs by up to five times while still trailing them on the agentic coding tasks developers care about most.