xAI · Tuesday, September 22, 2026
xAI launches Grok 4.7 for coding and knowledge work
The File desk · Sep 22, 2026, 10:25 AM UTC
Status
Confirmed (xAI news post, Sept. 21, 2026)
On Monday, Sept. 21, 2026, xAI published Introducing Grok 4.7. The company calls it its most powerful model for coding and knowledge work — a new, larger base than Grok 4.6, trained with a longer reinforcement-learning run on harder multi-hour tasks, with stronger self-checking and longer-context handling. xAI says it is served at the same price and speed as Grok 4.6, starting at $2 per million input tokens and $6 per million output tokens, with a fast variant at twice the output speed and twice the price. The post puts Grok 4.7 at 46.3% on CursorBench 4.0 (xHigh) and says availability starts in Cursor, Grok Build, the Grok API, third-party coding harnesses, and cloud routers. xAI also describes a new safeguard stack and a 62.4% score on LatchBio’s biosafety benchmark. No separate system-card PDF was linked on the launch page the desk fetched.
On Monday, Sept. 21, 2026, xAI published Introducing Grok 4.7.
The company calls it its most powerful model for coding and knowledge work — a new, larger base than Grok 4.6, trained with a longer reinforcement-learning run on harder multi-hour tasks, with stronger self-checking and longer-context handling.
xAI says it is served at the same price and speed as Grok 4.6, starting at $2 per million input tokens and $6 per million output tokens, with a fast variant at twice the output speed and twice the price.
The post puts Grok 4.7 at 46.3% on CursorBench 4.0 (xHigh).
Availability starts in Cursor, Grok Build, the Grok API, third-party coding harnesses, and cloud routers, the company says.
xAI also describes a new safeguard stack and a 62.4% score on LatchBio’s biosafety benchmark.
No separate system-card PDF was linked on the launch page the desk fetched.
What is still unknown or disputed
- Full system card / model card PDF if published separately; exact API model ID string in docs.x.ai beyond the marketing name Grok 4.7; independent replication of benchmark tables.
Primary sources
Every claim in this story is drawn from the documents below. If a fetch failed, that is recorded on the card.
Source 1
Introducing Grok 4.7
xAI / SpaceXAI · September 21, 2026
Grok 4.7 is our most capable model for coding and knowledge work. It works longer on difficult tasks, checks its own work more carefully, and comes with our best-calibrated safeguards to date. Served at the same price and speed as Grok 4.6, it is highly competitive in its class.
https://x.ai/news/grok-4-7