Gemini Omni 1.2 Flash Pricing: Cost per Second and 360p to 4K

👁 114 reads

A second of 720p video with audio from Gemini Omni 1.2 Flash costs $0.0695. That is exactly what I paid for each second generated on November 5, 2026, the model’s launch day, and it matches Google’s official pricing sheet. That works out to roughly $0.70 for a 10-second clip and $1.04 for the maximum 15-second generation.

Google promised a 30 percent price cut compared to Omni 1.1, and that held true in practice: the exact same 6-second clip in 720p cost me $0.61 on version 1.1 and $0.42 on 1.2. Below is the full resolution breakdown, a comparison with rival video generators, and several pitfalls that only became obvious once the invoice arrived.

How Google calculates pricing

Omni does not charge a flat rate per clip. Instead, you pay for video output tokens at $12 per million. According to Google’s documentation, one second of 720p video equals 5,792 tokens, giving the base rate of $0.0695. Higher resolutions rely on upsampling from the 720p base and are billed at higher rates. There is no free tier available in the API.

These dollar rates match what OpenRouter billed me and align with Google’s pricing documentation. Audio generation is included in the price, and both 16:9 and 9:16 aspect ratios cost the same.

The resolution test: the same jar across four tiers

To test generation across resolutions, I prompted an advertisement for pickled cucumbers in 360p, 720p, 1080p, and 4K, using the exact same prompt and the same seed.

Here is an important detail omitted from the initial launch announcements: changing the resolution parameter does not simply upscale the existing clip. Even with an identical seed, each resolution setting produced an entirely different scene. At 1080p, a hand entered the frame; at 360p, it did not. If you want to follow Google’s recommended workflow-drafting in 360p and upscaling the approved version to 4K-you must use the dedicated upscaling endpoint in the Gemini API or Vertex AI. Merely modifying the resolution parameter will generate a new clip from scratch.

Omni 1.2 vs competitor pricing

Here are the real costs per second of 720p video with audio that I paid during tests with 6-second clips on November 5, 2026:

For additional context from OpenRouter’s price list: Veo 3.1 Fast runs at $0.10 per second for 720p with audio, full Veo 3.1 costs $0.40 per second, and Seedance 2.5 in 720p comes out to around $0.23 per second. That makes Omni 1.2 the second-cheapest option overall, surpassed only by Veo 3.1 Lite. However, Lite only generates clips of 4, 6, or 8 seconds, and in my tests it refused to execute requested camera movements. You can read a detailed quality breakdown in my comparison guide, Omni 1.2 vs Seedance, Kling, Veo, and Wan.

Total test run spend

Across my evaluation, I generated 24 clips ranging between 4 and 15 seconds: 14 clips using Omni 1.2 (including the resolution scaling ladder) and 10 across five competing models. My total spend was $13.02. Nine requests were rejected outright due to content filtering or region restrictions on video editing in the EEA, Switzerland, and the UK; none of those rejected requests incurred charges.

Three ways to avoid overpaying

• Draft in 360p. It is three times cheaper than 720p and generates twice as fast. It provides more than enough clarity to evaluate framing, camera path, and pacing.
• Do not order 4K by default. One second of 4K equals the cost of three seconds of 720p. For short-form social formats like Instagram Reels and TikTok, 720p or 1080p is more than sufficient.
• Size clip length to spoken lines. You pay for every second, but making a clip too short risks clipped speech. In my tests, fitting two brief spoken lines required an 8-second generation; at 6 seconds, the model cut off half a sentence.

The older version 1.1 will not vanish immediately. Beginning December 3, 2026, Google will automatically route API requests to 1.2 and bill them at the lower rates, meaning existing production code targeting 1.1 will automatically receive the price reduction without any manual code changes. For broader architecture details, see the Gemini Omni Flash overview hub.

// read moreGemini Omni 1.2 Conversational Video Editing: How It WorksAI ToolsGemini Omni 1.2 Flash Prompting Guide: 7 Rules and 9 PromptsGuidesGemini Omni Flash: Video Quality, Pricing, and Test ResultsAI Tools

Kinetyra