Google Updates Gemini Omni Flash Preview
Google updated its Gemini Omni Flash Preview documentation on August 27, 2026, describing a multimodal model for video, image, and text tasks that can generate video and text in one model. Nokia Power User reports that a "Gemini Omni Flash 1.1" rollout has begun after Pre-GA delays, but Google's retrieved documentation identifies the available model as gemini-omni-flash-preview and retains Preview status.
Google updated documentation for Gemini Omni Flash Preview on August 27, 2026. The Google Cloud page describes it as a multimodal model for video, image, and text tasks, optimized for video generation and able to produce video alongside text responses in a single model.
A separate August 27 report from Nokia Power User describes a rollout under the name "Gemini Omni Flash 1.1" following minor Pre-GA delays. However, the retrieved Google documentation identifies the model ID as gemini-omni-flash-preview, not a 1.1 version, and labels the offering Preview under Google's Pre-GA terms. Google has not provided a retrieved release note confirming the "1.1" naming or the reported delay.
Documented capabilities and limits
According to Google's documentation, the model accepts text, image, and video inputs, while audio input is listed as unsupported. Its documented limits are 131,072 maximum input tokens and 57,920 maximum output tokens. The capabilities list indicates support for speech, music, and sound effects in generated output.
The documentation also states that the preview offering may be used for production or commercial purposes, including disclosure of generated output to third parties, subject to the applicable Google Cloud agreement and Pre-GA terms. Use requires a Google Cloud project with billing and the Agent Platform API enabled for the documented example-app deployment flow.
What remains unverified
The Nokia Power User report attributes the alleged rollout to Google Cloud and Pre-GA availability and claims speed, latency, and cost improvements. The retrieved Google Cloud page does not provide comparative latency, pricing, benchmark, or version-change figures. Developers evaluating the model should therefore treat those performance claims as unverified pending a Google release note, API changelog, or updated model card.
For teams building conversational video workflows, the material distinction is that Google's current documentation describes a preview, video-output-capable model with a large input context window. Across comparable generative-video preview releases, production evaluation commonly requires separate measurement of generation latency, output consistency, safety behavior, quota availability, and per-request cost because those properties are not established by a model's modality list alone.
Key Points
- 1Google documents Gemini Omni Flash Preview for text, image, and video inputs, with video and text outputs from one model.
- 2The documented model ID is gemini-omni-flash-preview; Google documentation retrieved here does not confirm a "1.1" release designation.
- 3Teams assessing generative-video previews commonly need independent tests for latency, cost, quotas, output quality, and safety behavior.
Scoring Rationale
A Google Cloud multimodal video-generation model is relevant to developers building media-generation and conversational editing workflows. The available first-party evidence documents a Preview offering, while the reported version-specific rollout and performance improvements lack confirmation in the retrieved Google material.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
