Grok 4.7 launches with long-task focus and unchanged $2/$6 API pricing
APPSO · wechat · 2026-09-22
xAI released Grok 4.7, positioned as its strongest coding and knowledge-work model, live in Cursor, Grok Build and Grok API. Built on a larger base model with much longer RL and harder training tasks, it scores 46.3% on CursorBench 4.0 xHigh (up from 40.4%), 71.0% on DeepSWE v1.1 and 38.0% on TerminalBench 4.0, though it still trails Fable 5.1 Max and GPT 5.6 Sol Max overall. Artificial Analysis gives it a 46 IntelligenceIndex, placing xAI among the top four labs.
- Pricing unchanged from 4.6: $2/M input, $6/M output tokens, with a 2x-speed, 2x-price tier
- xAI pushes professional work: a Box insurance demo found $82K duplicate invoices and a $143K deductible issue in a $2M claim; new safety stack scores 62.4% on LatchBio biosecurity, with only 3.3% of dangerous dual-use prompts passing HackerBench v0.3
- Community reception is split: impressive cheap Three.js/web demos ($0.20 vs $2.05 for Claude Opus 5) alongside complaints about stiff physics, unstable prompt understanding and fast token burn
- Roadmap: Grok 4.8 (2.5T params, new C++ training stack), with Grok 5 framed as an AGI bet
More from coding & agent
- How to cut Fable 5.2 token costs by 90%: Arc routing plus a lean CLAUDE.md — EXM7777 · 2026-11-12
- Hamel Husain's complete AI evals framework for PMs and engineers — aakashgupta · 2026-11-12
- AI coding in one year: from unusable to full automation with no code review — angelaqdai · 2026-11-12
- Interactions Tool Searches Web Animations by Motion for Claude Code to Rebuild — yuwen_lu_ · 2026-11-12
- Agent Substrate Calculator Estimates Session Resume Speed Under Load — pjausovec · 2026-11-12
- Senate Bill Would Force Big Platforms to Accept AI Agents, Raising Auth Privacy Questions — Prof_CarlaReyes · 2026-11-12