Grok 4.7 Launches: Built for Long Coding Tasks, 2x Speed at Half the Price
机器之心 · wechat · 2026-09-22
xAI released Grok 4.7, positioned as its best model for coding and knowledge work at twice the speed and half the price of comparable models. Key points:
- Uses a larger base model with extended RL training on harder, multi-hour tasks, improving self-checking and long-context management.
- Long-task benchmarks improved sharply: CursorBench4.0 46.3% (vs 40.4%), DeepSWE v1.1 71.0%, Terminal-Bench4.0 up from 20.3% to 38.0%.
- Professional work scores also rose: AABriefcase v1.1 1657 (vs 1546), GDPval Elo 1605→1695, plus gains in legal, medical and engineering benchmarks.
- New safety stack: strongest refusals and jailbreak resistance in internal testing; only 3.3% of high-risk dual-use prompts allowed on HackerBench v0.3; invite-only red-teaming for cybersecurity partners.
- Pricing unchanged at $2/M input and $6/M output tokens; available via Cursor, GrokBuild and the Grok API.
The report frames this as frontier models moving toward "completing entire tasks," though online reaction has been mixed.
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