Google

Recently shipped Gemini Omni Flash GA, dedicated transcription models, and Workspace data import for Teams and OneDrive.

Gemini Omni Flash reached GA for video generation and editing. gemini-omni-1.1-flash extends existing clips, interpolates between two images, and supports up to 4K output via upscaling. The gemini-omni-flash-preview endpoint deprecates September 30, 2026.

Dedicated speech-to-text models went GA. Gemini 3.5 Transcribe handles non-streaming transcription across 85+ languages with diarization, word-level timestamps, and custom vocabulary. Gemini 3.5 Transcribe Live adds bidirectional streaming over WebSockets with interim events and multiple VAD strategies.

Workspace data import now covers Microsoft Teams and OneDrive. Admins can migrate channel messages, group chats, and DMs from Teams, plus large-scale OneDrive file migrations with permissions, at no additional tool cost. Calendar also improved third-party video meeting joins by placing conferencing URLs in the Location field for external recipients and adding a structured Join button for incoming invites.

Workspace admin and collaboration controls expanded. Chat spaces gained a View members setting with four permission levels, and Classroom now supports Context-Aware Access policies. Sheets preserves grouped pivot fields across Excel import/export, and Meet hardware touch controllers gained direct Take notes for me control plus a refreshed UI on Neat and Poly devices.

Gemini app now generates interactive simulations. Responses can include rotatable 3D models, physics simulations, and interactive tables built from a single prompt. The JS and Python GenAI SDKs added audio transcription config modes, environment file support, and JPEG 2000 video content; the Python SDK also made the deferred service tier available on Vertex. Chrome 153 reached early stable on desktop and Android, with 152 stable rolling out; ChromeOS LTS-144 and M-151 stable received security fixes including critical WebGL and Chromoting CVEs.