Unintelligible output text from Cursor Grok 4.6

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

I use Cursor Grok 4.6 by typing in Italian in the chat and having Grok 4.6 respond in Italian. Unfortunately, the sentences the AI generates are difficult to understand; the words are in Italian, but the meaning is far from clear, and I often have to ask it to explain exactly what it means. This always happens with Grok, but with other AIs—such as Claude Sonnet or Opus—it never happens: the meaning is always understandable. Unfortunately, this means I don’t use Grok but rely on other AIs instead, which results in a significant increase in costs.
I’m asking the team that manages Grok to improve the clarity of the responses.
P.S. The same thing happens with Composer 2.5 (which is often incomprehensible).

Steps to Reproduce

Submit a request to Grok

Expected Behavior

May the meaning of the answer be clear

Operating System

Windows 10/11

Version Information

Version: 3.20.17 (user setup)
VS Code Extension API: 1.128.0
Commit: 0c32194e3fb5ffaced9fb36430b860ec301e1fc0
Date: 2026-09-12T03:16:10.634Z
Layout: IDE
Build Type: Stable
Release Track: Default
Electron: 42.10.0
Chromium: 148.0.7778.280
Node.js: 24.18.1
V8: 14.8.178.38-electron.0
xterm.js: 6.1.0-beta.291
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use?

Cursor Grok 4.6, Composer 2.5

For AI issues: add Request ID with privacy disabled

8ffba6e0-ea3c-4285-bf76-52a56319f2b8

Does this stop you from using Cursor

No - Cursor works, but with this issue

Hey @brales, thanks for the detailed report.

I checked the Request ID you attached. The request itself completed normally, so this is not a failure. It is about how Cursor Grok 4.6 writes. Grok 4.6 has a very compact style with short broken phrases and fewer connecting words, and in Italian it is harder to read than in English. Claude usually writes fuller sentences by default, so its answers feel clearer. I passed this to the team that works on model output quality.

For now, the best lever is a persistent instruction via a User Rule. It fixes most of the problem:

  1. Open Cursor Settings with Ctrl+Shift+J on Windows.
  2. Rules → User Rules → Add Rule.
  3. Paste and save:
Rispondi sempre in italiano corretto e naturale, con frasi complete e ben collegate tra loro.
Non usare frammenti telegrafici, abbreviazioni o gergo non spiegato: se usi un termine tecnico o una sigla, spiegalo la prima volta.
Riassumi sempre in linguaggio semplice cosa hai fatto e cosa hai trovato, come se il lettore non avesse visto il codice.
  1. Open a new chat with Cursor Grok 4.6 and ask your usual question. The same instruction also works for Composer 2.5.

Is the answer easier to understand now? If you still get an unclear answer, send the answer text here or a screenshot, plus its Request ID via the three dots on the message → Copy Request ID. Real examples like that are exactly what the team needs to improve this.

Thanks for your reply, Dean. I made the change you suggested. I’ll use Grok for a few days and then let you know if the results are clearer.
Bye