The launch
On September 21, 2026, xAI (which now signs its official pages as SpaceXAI) released Grok 4.7, described as its most capable model for coding, agentic tasks and knowledge work. The company says it "works longer on difficult tasks, checks its own work more carefully" and has its best-calibrated safeguards yet, at the same price and speed as Grok 4.6.
What changed
- A new, larger base model than Grok 4.6, with a longer reinforcement-learning run weighted toward problems that take many hours.
- Better at verifying its own work and managing long context.
- Trained to natively understand the Grok Bot harness, which xAI says improves conversational and general knowledge work.
- An entirely new safeguard stack: xAI says it is the strongest model it has tested on refusals and jailbreak resistance. On its internal HackerBench v0.3, it let only 3.3% of risky cyber prompts through. Select security partners got invite-only access to red-team capabilities.
Pricing and specs (xAI docs)
- API ID:
grok-4.7; 500k-token context window; text and image input, text-only output. - Below 200k prompt tokens: $2 input / $0.50 cached input / $6 output per million. Above: $4 / $1 / $12.
- Reasoning effort: low, medium, high (default) and xhigh.
- Grok 4.7 Fast: same model, twice the speed at twice the price, only in Cursor and Grok Build.
Performance
xAI and press reports cite 46.3% on CursorBench 4.0 and 71% on DeepSWE 1.1. Decrypt notes it came second on some tests: 1695 on GDPval versus 1735 for Claude Fable 5.1, and behind GPT-6 Astra on an electrical-engineering benchmark. Decrypt also reports the launch followed at least five delays since late July.
How to use it
Available in the Grok app, Cursor, Grok Build, the xAI API and other coding harnesses and model routers. For Grok Build in the terminal, Neowin cites the official installer: curl -fsSL https://x.ai/cli/install.sh | bash.
Heads-up: X Search billing change
On the same day, xAI changed pricing for the API's X Search tool: instead of $5 per 1,000 calls, it now charges $5 per 1,000 posts and $10 per 1,000 profiles returned, according to Search Engine Watch. Brand-monitoring apps that pull many posts per search may see costs rise sharply.
What it means for businesses
Grok 4.7 is a low-cost option for coding agents and long documents. Test it on your real tasks and compare cost per completed task, not just price per token. If you use X Search, redo your cost math.
Sources
- xAI, "Introducing Grok 4.7": https://x.ai/news/grok-4-7
- API release notes: https://docs.x.ai/developers/release-notes
- Decrypt: https://decrypt.co/378824/xai-launches-grok-4-7
- Neowin: https://www.neowin.net/news/grok-47-brings-multi-hour-coding-and-2-input-pricing/
- Search Engine Watch: https://searchenginewatch.com/grok-4-7-launches-as-xai-switches-x-search-to-per-post-billing/
Want to pick the right AI model for your team? KGA IT offers a free consultation: kga-it.com.