o4-mini-deep-research launched on 26 June 2025 alongside o3-deep-research, as the affordable half of OpenAI's research pair. It performed the same job — searching the web, running code and pulling private material through MCP connectors before writing a synthesised, cited report — at 2 dollars per million input tokens, 50 cents for cached input and 8 dollars for output. That is a fifth of the larger model's rate, and identical to what plain o3 cost after its June 2025 price cut.
The rest of the specification matched its twin: a 200,000-token context window, up to 100,000 output tokens, text and image input, text output, and a knowledge cutoff of June 2024. Rate limits were more generous than o3-deep-research at every tier, which fits the intended use — running many research jobs rather than a few careful ones.
Both models were deprecated on 22 April 2026 and shut down together on 23 July 2026, under the three-month notice period OpenAI applies to specialised variants rather than the six months given to general models. The recommended replacement is gpt-5.6-sol.