Google announced multiple feature updates to its Gemini app in April 2026, according to a company blog post published 24 April [source]. The changes include expanded multimodal capabilities, new integration points with Google Workspace, and adjustments to context window handling for certain model tiers.

The update introduces what Google calls "Gems 2.0," allowing users to create persistent custom instructions that apply across conversations. The company states these instructions now support image and document uploads as part of the configuration, a capability not present in the original Gems feature launched in 2024.

Google also announced that Gemini Advanced subscribers will gain access to a "Deep Research" mode, described as an extended reasoning process for complex queries requiring synthesis across multiple sources. The feature generates intermediate research plans visible to users before producing final outputs. No technical details about the underlying model changes were provided.

The post confirms that Gemini 1.5 Pro remains the default model for Advanced tier users, with Gemini 2.0 Flash available as an optional selection. Google states that context window limits for uploaded files have increased to 2 million tokens for 1.5 Pro users, up from the previous 1 million token limit.

Additional changes include a redesigned mobile interface with persistent chat history access and new voice interaction modes supporting 40 languages. The company did not specify whether these voice capabilities rely on existing speech models or represent new architecture.

The updates began rolling out to users on 24 April, with full availability expected within two weeks according to the post. Google provided no information about changes to safety filters, output formatting, or underlying training data.