o4-mini-deep-research

OpenAI · o-series

Lower-cost OpenAI deep research model for source-heavy investigations when throughput and budget matter.

Type
language
Context
200K tokens
Max Output
100K tokens
Status
current
Input
$2/1M tok
Output
$8/1M tok
API Access
Yes
License
proprietary
deep-research reasoning web-search mcp file-search cost-efficient
Released June 2025 · Updated April 8, 2026

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on April 8, 2026.

o4-mini-deep-research is OpenAI’s faster, lower-cost deep research model. OpenAI positions it as the affordable route for complex, multi-step research tasks that still need web search, MCP-based private-data access, and longer-form source synthesis.

This is a useful model when you want the deep research workflow shape without paying the premium attached to o3-deep-research on every run.

Capabilities

Like the higher-end o3-deep-research route, this model is designed to search and synthesize information from the web and from your own data through MCP connectors or vector-store file search. It is optimized for building comprehensive reports, not just answering one question from memory.

That makes it a practical fit for recurring research pipelines, analyst-assist workflows, internal briefs, market scans, policy watchlists, and other repeatable jobs where cost sensitivity is real but a normal chat model would be too shallow.

Technical Details

Current published limits:

  • Context window: 200,000 tokens
  • Max output: 100,000 tokens

The model supports text input/output and image input. Streaming is supported, but function calling and structured outputs are not. As with the rest of OpenAI’s current deep research stack, the model expects at least one data source such as web search, remote MCP servers, or file search. Optional code interpreter support can be added for heavier analysis.

Operationally, it behaves more like an agentic research worker than like a generic assistant turn.

Pricing & Access

Published pricing (per 1M tokens):

  • Input: $2.00
  • Output: $8.00

That is much cheaper than o3-deep-research, which is why this route makes sense for higher-throughput research programs or first-pass investigations. Access is through OpenAI API deep research workflows, and background mode is still the sensible default for longer jobs.

Best Use Cases

Use o4-mini-deep-research for repeated research tasks where budget and throughput matter: weekly market briefs, evidence gathering, connector-backed knowledge checks, and internal reporting that still needs real source synthesis. When report quality and reasoning depth matter more than cost, escalate to o3-deep-research.

Comparisons

  • o3-deep-research (OpenAI): Higher-capability premium option for the hardest research tasks.
  • o4-mini (OpenAI): Better for fast reasoning without the full deep research workflow.
  • GPT-5 mini (OpenAI): Better cost-sensitive general assistant route, but not a direct replacement for deep research behavior.