Gemini Omni 1.2 Flash Review: Video Generation Model Capabilities, Pricing and API Access

Gemini Omni 1.2 Flash is Google's generally available multimodal model for video generation and editing. Released on November 5, 2026, it produces short clips for ads, product videos and social content with synchronized audio. The model ID is gemini-omni-1.2-flash and it succeeds the 1.1 version released earlier in 2026.

The service runs through the Gemini API, Google AI Studio, Vertex AI and consumer tools such as Google Vids and Flow. Users supply text prompts, up to six reference images, short motion clips or existing video for editing. Output clips reach 15 seconds at 24 fps in 360p, 720p, 1080p or 4K after upscaling.

Gemini Omni 1.2 Flash Capabilities and Workflow

The model creates native clips between 3 and 15 seconds with first- and last-frame interpolation. It accepts up to three seconds of motion reference and supports multi-turn conversational edits that let users refine background, lighting or product details without regenerating the full clip. Audio generation includes dialogue, sound effects and music, with improved lip-sync for multiple speakers.

Reference adherence covers consistent characters, typography and branding across shots. Aspect ratios are limited to 16:9 and 9:16. Draft generation at 360p or 720p followed by selective upscaling helps control costs. Generation times range from 20 to 136 seconds depending on resolution and complexity, according to independent tests published in November 2026.

Outputs carry SynthID watermarks. The model is positioned as a cost-effective Flash-tier option focused on instruction accuracy rather than maximum photorealism. Limitations remain in fast action sequences and complex physics.

Gemini Omni 1.2 Flash Pricing, Access and Comparisons

API pricing stands at approximately $0.07 per second of 720p output, 30 percent lower than the prior version. A 15-second 720p clip costs roughly $1.04 before additional input charges. 360p drafts run at about $0.023–$0.026 per second. Token rates are listed near $12 per million video output tokens plus $1.50 per million input tokens. Upscales to 1080p or 4K incur extra per-second fees. No free API tier is available.

Access is generally available via Gemini API and Vertex AI with enterprise SLAs. Google AI subscribers can generate 15-second clips in the Gemini app. Traffic from version 1.1 is scheduled to route to 1.2 rates on December 3, 2026.

Compared with Veo 3.1, the Omni 1.2 Flash model offers faster turnaround and lower cost for text-heavy or editing tasks but trails in cinematic quality. ByteDance Seedance 2.5 supports clips up to 30 seconds and stronger motion realism, while Runway Gen-4.5 and Kling 3.0 provide different length and fidelity trade-offs. The following table summarizes key differences.

Public user feedback highlights reliable instruction following for explainers and typography but notes occasional struggles with rapid camera moves. Complaints center on pricing for high-resolution upscales and limits on complex action. No widespread reports of data leakage have surfaced since launch.

Limitations and Final Assessment

Documented constraints include restricted fast-action handling and the need for allowlisted access for certain audio references. Higher resolutions increase costs without guaranteed quality gains in every case. The model does not replace higher-fidelity tools when maximum realism is required.

Overall, Gemini Omni 1.2 Flash delivers measurable improvements in prompt adherence and editing speed at a reduced price point. Its conversational workflow and native audio support suit marketing and explanatory content, while users seeking longer or more photorealistic results continue to evaluate alternatives.