Grok 4.7 Costs the Same as 4.6 but Handles Longer Coding Tasks

xAI launched Grok 4.7 on 21 September 2026, its most capable model yet for coding and knowledge work. The company says it works longer on hard problems, checks its own output more carefully, and costs the same as its predecessor while beating pricier rivals on price, a shift worth noting if your business already uses AI tools.

What did xAI announce with Grok 4.7?

Grok 4.7 runs on a larger base model than Grok 4.6, trained with a longer reinforcement learning run on harder, longer running tasks. xAI's own 21 September 2026 announcement says the model checks its own work more carefully and manages longer context better than its predecessor. On CursorBench 4.0, a benchmark built around longer coding tasks, Grok 4.7 scored 46.3 percent against Grok 4.6's 40.4 percent. Pricing stays at 2 US dollars per million input tokens and 6 US dollars per million output tokens, unchanged from Grok 4.6, which xAI says makes it highly competitive against pricier rivals in its class.

Weeks before the launch, xAI founder Elon Musk had already set expectations for how the model would compare to rivals such as Anthropic.

Grok 4.7 will exceed all current models. That said, Anthropic is a great company and will probably release improved models soon. However, the SpaceX training corpus is so awesome & unique that I would be shocked if any model is better at real-world engineering than 4.7.from Elon Musk, founder, xAI (X, 12 August 2026)

What can Grok 4.7 actually do?

• Handle longer, harder coding tasks with less need for a human to check its work along the way.
• Build documents, presentations and interfaces that xAI describes as more tasteful and higher quality than Grok 4.6's output.
• Run through the Grok API, Cursor, Grok Build, and third party coding tools and cloud platforms, available the same day it launched.
• Resist jailbreak attempts better than any previous xAI model, scoring 62.4 percent on LatchBio's biosafety benchmark while still allowing legitimate security work through.

Does this matter if you do not write code?

Most businesses are not going to build software with Grok 4.7 directly, but the price war behind it does matter. xAI, OpenAI, Anthropic and Google are all racing to cut the cost of frontier AI while raising what it can do, and that competition is what makes tools like ChatGPT, Claude and Gemini cheaper and more capable for an ordinary business user every few months. If your team already uses AI for reports, quotations or customer replies, this is the kind of release that quietly makes next month's version of the same tool better without you having to switch anything or pay more for it.

Also today

• Xiaomi open sourced its MiMo V2.6 model family on 22 September 2026, including a trillion parameter Pro model it says performs on par with Claude Opus 5 on most agent benchmarks (Xiaomi MiMo).
• Malaysia's Digital Ministry completed its draft AI Governance Bill and is preparing it for Cabinet, with Digital Minister Gobind Singh Deo saying it could reach Parliament by the third quarter of 2026 (Bernama, via TechRepublic).
• Moonshot AI rolled out Kimi K2.8 Preview inside Kimi Code and Kimi Work on 11 September 2026, a mid tier model the company says performs close to its flagship Kimi K3 (Pandaily).

What to do this week

You do not need to touch Grok 4.7 yourself to benefit from this race. What matters is picking one AI tool for one recurring task at work, whether that is drafting quotations, replying to customers, or building a small internal tool, and using it consistently instead of waiting for the next flashy model name. That is exactly what we walk through, hands on, in our AI Full-Stack Building for Business program (also known as vibe coding), where you build a real tool for your own business instead of just reading about what AI can do.