Model release

Deep Research Max Preview: April 2026 Release

Google released Deep Research Max Preview on April 21, 2026. This update adds collaborative planning, visualization support, MCP server integration, and File…

Released Published 3 min readModel evidence ↗
Deep Research Max Preview (Apr-21-2026) release illustration — Google

Official Announcement Details

Google officially released the Deep Research Max Preview on April 21, 2026. The specific model identifier provided in the official changelog is deep-research-max-preview-04-2026. This release belongs to the Gemini family and supports text input and text output modalities. The source documents state that this version offers maximum comprehensiveness for automated context gathering and synthesis.

Documented Features and Specifications

The release includes several distinct capabilities listed in the official documentation. These include collaborative planning, visualization support, MCP server integration, and File Search. The model has a documented context window of 131,072 tokens and a maximum output token limit of 65,536. Open weights are not confirmed for this release. Pricing information is currently null in the available data, meaning no specific cost structure is established in the source facts.

  • Collaborative planning capability
  • Visualization support
  • MCP server integration
  • File Search functionality
  • Context window: 131,072 tokens
  • Max output tokens: 65,536

Evaluation Procedure for Engineers

Engineers can evaluate this model by designing tests that validate its specific documented features. First, test the collaborative planning feature by providing a complex multi-step research prompt and measure the coherence of the generated plan. Second, validate the MCP server integration by connecting a local MCP server and comparing the success rate of tool calls against a baseline. Third, measure the effectiveness of File Search by uploading a diverse set of documents and comparing retrieval accuracy against previous manual search methods. Finally, compare the visualization outputs by generating charts from raw data and assessing their readability and accuracy against standard plotting libraries. These steps allow for a practical assessment of the new capabilities without assuming unverified performance improvements.

Primary sources

QUICK ANSWERS

Frequently asked questions

What is the exact date of the announcement?

The original announcement date recorded in the source facts is April 21, 2026. This historical date distinguishes the release event from the current publication date of this article.

Are open weights available for this model?

The source facts indicate that open weights are not confirmed (openWeights=false). This does not establish that local installation is impossible or that API access is mandatory; it only records the lack of confirmation regarding open weight availability.