
Understanding Microsoft Copilot’s Evolving AI Capabilities
Microsoft Copilot, while not universally praised, effectively provides some free features akin to ChatGPT, albeit with built-in restrictions and censorship. Currently, the platform’s “Think Deeper”functionality utilizes the o3-mini-high model, which has a knowledge cutoff set in October 2023. However, Microsoft seems to be experimenting with the o4-mini-high model through an A/B testing approach.
Modes of Operation in Copilot
Copilot offers users a choice between two default modes, with a third option, Deep Research, available for those subscribing to Copilot Pro for $20. The first mode, Quick Response, taps into one of OpenAI’s standard models, while the “Think Deeper”mode engages a more advanced reasoning model. As confirmed in March 2025, the “Think Deeper”feature activates the o3-mini-high model, a premium offering from OpenAI.
“Think Deeper’s reasoning model has been upgraded to OpenAI’s o3-mini (high).This means you’ll wait less to get answers, ” Microsoft noted in the March 2025 release notes.
Transition to Newer Models
Despite its initial offer with ChatGPT Plus and Teams/Enterprise, the o3-mini-high model has reached its end and has been superseded by the more powerful o4-mini-high model. There are indications that this new model is being integrated into Copilot, specifically activated through the “Think Deeper”function.
Insights from Recent Testing
In assessments conducted by Windows Latest, it was noted that Microsoft intermittently replaced the existing o3-mini-high model with the newer o4-mini-high model during testing sessions. This change came to light when users sought to clarify the knowledge cutoff date of Copilot’s responses.

When utilizing a Microsoft account, further testing revealed that with “Think Deeper”enabled, Copilot indicated that its training data extended to June 2024:

Model Knowledge Cutoff Dates
The knowledge cutoff dates for various OpenAI models are as follows:
- o3-mini / o3-mini-high: October 1, 2023
- o3: June 1, 2024
- o4-mini / o4-mini-high: June 1, 2024
- GPT-4.1 family (GPT-4.1, GPT-4.1 mini, GPT-4.1 nano): June 2024
There is speculation regarding whether “Think Deeper”might employ the o3 model, which also concludes training in June 2024. However, implementing such an expensive model for the quick-response framework of “Think Deeper”seems unlikely, yet it cannot be completely discounted.
Leave a Reply ▼