Blog

Grok 4.7 builds on 4.6. Migration needs attention to the 200k-token pricing threshold, caching and preservation of encrypted reasoning history.

A new Czech test of local voice cloning exposes a practical problem: a model can sound excellent in English and fail on Czech or code-switching. Companies need to evaluate supported languages, weight licences, latency, and consent—not only voice similarity.

Grok Build is no longer only a terminal coding agent. Xai open-sourced it and added workflows that split large jobs across agents, verify results, and save the process in the repository. The practical move is to start with bounded audits, budgets, and human approval.