Grok 4.7 targets longer coding and knowledge-work tasks
The September model release pairs a larger base model with a longer reinforcement-learning run.
THE STORY IN THREE POINTS
- Grok 4.7 launched on September 21.
- Access includes Cursor, Grok Build and the API.
- The launch lists starting token prices of $2 input and $6 output per million.
The announcement
The Grok team introduced Grok 4.7 on September 21 for coding and knowledge work. It describes a larger base model than 4.6 and longer reinforcement learning on harder tasks. The model is available through Cursor, Grok Build and the API, with starting prices of $2 per million input tokens and $6 per million output tokens. Benchmark and safeguard improvements are attributed to the official announcement.
Editorial context
Longer work demands more than maintaining a conversation. An assistant must preserve the objective, recognize when a result is wrong and leave something another person can inspect. For a software task, that might mean passing tests and documenting unresolved behavior. For an office task, it might mean a usable report with claims traceable to the supplied material.
Editorial context: what to watch
A practical comparison could use the same repository and documents across model versions, counting completed tasks and review time. The announced prices should be considered together with token consumption and retries. Publishing failures as well as successful examples would clarify whether stronger performance on lengthy benchmarks transfers to the particular work a reader wants to delegate.
Sources: Official Grok 4.7 announcement, September 21 ↗
