A user can select various models for a deep research request. Does that affect the model that executes the deep research work? What model(s) do the deep research task?
#Model used for deep research
sterile scaffold
pallid atlasBOT
Hi @sterile scaffold. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
The knowledge sources indicate that deep research in ChatGPT is powered by a specialized model, specifically the OpenAI o3 reasoning model. Deep research is fine-tuned on o3 and is designed for in-depth, multi-step research using data from the public web. This model autonomously searches, reads, and synthesizes information to produce thorough, cited reports on complex topics.
While users can select different models for general tasks in ChatGPT, the deep research capability itself is explicitly tied to the o3 model. There is no indication in the knowledge sources that selecting a different model (such as GPT-4o or o4-mini) will change the model that actually executes the deep research task. Instead, deep research always uses the o3 model for its main research and synthesis work. If you exceed your monthly limit, a lightweight, cost-effective version is used, but the documentation does not specify which model powers this fallback version—only that it is not the standard o3 model.
In summary:
- Deep research is executed by the o3 reasoning model, regardless of the model you select for other tasks.
- Model selection in the UI does not affect which model performs the deep research itself.
- If you hit your monthly limit, a lightweight version is used, but its underlying model is not specified in the available documentation.
References: