Grok 4.7 ships at Chinese-model prices and benchmarks well behind GPT-6 and Claude Fable 5.1

Elon Musk's xAI has released Grok 4.7, which it calls its most capable model yet for coding, agentic tasks and knowledge work. The company lists four changes over Grok 4.6: a new and larger base model rather than a reuse of the old one, a longer reinforcement learning run weighted toward problems that take many hours, better self-verification and long-context handling, and native support for the Grok Bot harness. Pricing stays at $2 per million input tokens and $6 per million output tokens, the same as Grok 4.6. The API lists a 500,000-token context window, a May 2026 knowledge cutoff, text and image input with text output, and reasoning effort settings from low to xhigh. It is callable through the xAI API, Cursor, Grok Build, OpenRouter, Vercel and Cloudflare. Independent scores put the release mid-pack. On the Artificial Analysis Intelligence Index v4.3.2, which combines ten benchmarks, Grok 4.7 scores 46 against 53 each for Claude Fable 5.1 and GPT-6. The gap widens in agentic coding: on Terminal-Bench 4.0 it reaches 26 percent, versus 60 percent for GPT-6 Astra and 55 percent for Claude Fable 5.1, with the cheaper DeepSeek V4.1 Flash slightly ahead at 27 percent. The pricing, closer to Chinese models than Western frontier models, reads as a response to that position.