Google Flow Pricing, Plans and Limits

In Google Flow a plan buys credits, and how many clips those credits make depends on the model: an 8-second Veo 3.1 Lite clip costs 10 credits, or 5 on Google AI Ultra, while Gemini Omni Flash 1.2 charges 3 credits a second at 720p and 1 at 360p on every plan, 24 or 8 for the same eight seconds. Everyone's 50 daily credits sit on top of each plan's monthly pool. Those two models are what Flow keeps: on December 5 Google retires Veo 3.1 Fast, Veo 3.1 Quality and Omni Flash 1.1.

Google Flow clips per month on each Google AI plan

The clip columns are ToolColumn arithmetic: the monthly pool divided by Google's credit cost per generation, one generation per clip and no retries. The 50 daily credits add five more Lite clips a day, ten on Ultra, or two 720p Omni Flash clips, and unused credits, daily or monthly, do not roll over. Without a plan, Google says video generation may be unavailable during peak hours, roughly 2 PM to 5 PM UTC.

Flow has no 5-hour or weekly window of its own: the credit balance is the limit. The compute-based limits that refresh every 5 hours up to a weekly cap belong to the Gemini app, where video generation also counts; Gemini limits by plan covers them.

Google Flow credit cost per generation by model

Google's Flow credits guide prices Omni Flash 1.2 by the second and Veo 3.1 Lite by the clip, and only Lite costs less on Ultra:

Ultra 20x subscribers and family plan managers can also make Lite clips for 0 credits in Lower Priority mode. Until December 5 Flow still runs the retiring models at their old rates: Veo 3.1 Fast at 20 credits a clip (10 on Ultra), Veo 3.1 Quality at 100, and Omni Flash 1.1 at 7, 10, 12 and 15 credits for 4, 6, 8 and 10 seconds at 720p. Veo Extend retired on November 5, so extending a clip now takes Omni Flash 1.2 and a paid plan.

Because Omni Flash 1.2 bills by the second, length drives its cost: a 10-second 720p clip costs 30 credits and a 15-second one 60, since seconds 11 to 15 cost double. On a paid plan, drafting at 360p and upscaling the keeper to 720p or 1080p for no credits costs a third of working at 720p throughout, 8 credits for eight seconds instead of 24; Google's models page lists that upscale for Plus, Pro and Ultra. Cost is not the only difference between the two models: Omni Flash edits existing clips, runs longer and is the only one that extends a clip, as Gemini Omni Flash vs Veo 3.1 sets out.

For Flow specifically, the current product help says Plus, Pro, and Ultra subscribers can purchase additional AI credits in supported regions except Japan. The general Google One purchase page currently gives a narrower Pro-and-Ultra eligibility statement, so check the signed-in checkout and region before treating a top-up as guaranteed. Model costs can change, so inspect the model selector before committing a project. The same AI credit balance also pays for Google Antigravity usage beyond its quota on Pro and Ultra, so heavy agent work can drain credits meant for Flow; Antigravity limits by plan explains its overage setting.

Budget retries by generation

Google prices Flow work per generation rather than per prompt. One request can create multiple outputs, so budget outputs and retries separately from prompts. A useful planning formula is:

shots × attempts per shot × outputs per request × credits per generation

For an illustrative eight-shot project with three attempts per shot and one output per request, the plan contains 24 generations. At Veo 3.1 Lite's 10 credits a clip, that is 240 credits, or 120 on Ultra; as eight-second Omni Flash 1.2 clips at 720p, 24 credits each, it is 576 credits on any plan. This is planning arithmetic, not a guaranteed finished-minute yield. Eight 8-second clips total 64 seconds of raw footage before rejected takes, extensions, edits, or upscaling. Google returns the credits of a generation that fails or hits a safety filter within an hour, but a completed result that is rejected still belongs in the retry budget.