Copilot Chat model picker swaps GPT names for Auto, Quick and Think Deeper
Reviewed by Helen Jones
Around 7 July 2026, Microsoft 365 Copilot Chat users saw the model picker stop listing named GPT versions such as GPT-5.5. In the desktop app and web, options shifted to Auto, Quick Response, Think Deeper, and in some tenants Opus (Claude). Microsoft appears to be steering people toward response style rather than model brand. Auto chooses an approach from the prompt. Quick Response favours speed for drafts and summaries. Think Deeper spends more time on complex analysis. By 8 July, named models reappeared for at least some users. Visibility of exact GPT labels and Claude options still varies by tenant and may keep changing.
Until now, many people treated the model picker like a quality dial. They picked a higher GPT number when a draft felt weak, then still rewrote the same vague prompt and blamed the model when the output stayed generic. The new labels make the real choice clearer: speed versus depth. That helps everyday users, but it also removes a false comfort. If named models come and go, the stable skill is matching the task to Auto, Quick, or Think Deeper, then checking facts, tone, and context before anything is shared.
Analysis
Treat this as a small playbook update, not a crisis. Open Copilot Chat once, note whether you see Auto, Quick Response, Think Deeper, Opus, or named GPTs, and write one line for your team: use Quick for short drafts and summaries, Think Deeper for multi-step analysis, and always review before send.
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Copilot Chat model picker swaps GPT names for Auto, Quick and Think Deeper", Collab365 Spaces. 2 sources referenced.