Google Flow: AI Filmmaking Studio with Veo, Nano Banana and Gemini Omni (2026)

Misplaced (2026-09-30)

This is an AI filmmaking/video note, not a coding-tool note; it belongs with the video models (see ai-video-generation-tools and google-veo in ai-models/video-models/). Not moved (rewrite-phase rule).

Google Flow: AI Filmmaking Studio with Veo, Nano Banana and Gemini Omni

Overview

Google Flow (introduced 2025-05-20, from Google Labs / Google DeepMind) is Google’s AI creative studio for images and video. In 2026 it is one unified workspace combining image models (nano-banana-2, nano-banana-pro, Nano Banana 2 Lite), video models (google-veo in Lite/Fast/Quality variants and the newer Gemini Omni Flash) and editing tools, with a Flow Agent in the prompt box.

Access: labs.google/fx/tools/flow, bundled into Google AI plans (not sold separately).
Models in Flow (support page, 2026-10-01): Veo 3.1 Lite / Fast / Quality; Gemini Omni Flash 1.1; Nano Banana Pro (Pro/Ultra), Nano Banana 2, Nano Banana 2 Lite (free).

What changed in 2026

• 2026-01-13: Veo 3.1 “Ingredients to Video” update: more expressive video, stronger character/background consistency, native 9:16 vertical output, and 1080p/4K upscaling (upscaling limited to Flow, Gemini API and Vertex AI).
• February 2026 (Flow update): Whisk and ImageFX merged into Flow (“generate, edit and animate everything in one unified workspace”); Nano Banana image generation built in and free; redesigned asset grid with collections; lasso selection plus natural-language edits (“remove the man”); Whisk/ImageFX users could migrate projects from March. Google cited 1.5 billion images and videos generated since launch.
• Nano Banana 2 (gemini-3.1-flash-image; release date 2026-02-26 per a secondary source) became the default image model in Flow; Nano Banana Pro (gemini-3-pro-image) remains stable for Pro/Ultra.
• I/O 2026 (May): Gemini Omni arrived in Flow for paid subscribers: video from text, images, audio and existing clips, video-to-video editing, 10-second clips, SynthID watermark; “Create custom voices” in some regions. The Gemini API video docs now recommend gemini-omni-1.1-flash as the default and keep Veo 3.1 (still listed as preview) for scene extension and last-frame control (per google-veo). Veo 4 is not confirmed by Google as of this check.
• Plans repriced: per RedShark News (secondary, I/O 2026 coverage), Google AI Ultra was cut in price and a higher Ultra tier added; Flow now draws on a monthly credit pool (see Pricing & Access).
• Imagen 4 is listed as shut down in the Gemini API docs (2026-10-02).

Problem It Solves

Traditional filmmaking needs script, assets, shooting, editing and grading. Flow lets one person generate consistent visual assets, iterate on scenes, and go from concept to a finished clip. Image models supply the “ingredients” (products, characters, backgrounds) that anchor video generation; Flow keeps them reusable via references (@charactername, visual ingredients, voice references) and start/end frames.

Architecture: Flow’s Creative Pipeline

Flow workspace (assets, collections, Flow Agent)  
   |-- Images: Nano Banana 2 / Lite / Pro (text-to-image, edits, lasso)  
   |-- Video:  Veo 3.1 (Lite/Fast/Quality) and Gemini Omni Flash  
   |        (text/frames/ingredients -> clip, extend, camera control, video-to-video)  
   '--> reference images and clips -> final video (audio, SynthID watermark)  

Image models (details)

• Nano Banana 2 / Lite: fast default models; Lite is the free option in Flow.
• Nano Banana Pro: premium model targeting “studio-quality 4K visuals, complex layouts, and precise text rendering” (Gemini API docs, 2026-10-02; see nano-banana-pro).

Flow workflow: from image to video

1. Create base images (product, lifestyle, environment) with Nano Banana.

2. Use them as ingredients/frames in Flow, with references for consistency.

3. Generate clips with Veo 3.1 or Gemini Omni Flash (Omni clips are currently 10 s); extend or edit clips (camera motion, add/remove objects, video-to-video on Omni).

4. Assemble, add audio/voice, upscale (1080p; 4K upscaling per the Jan 2026 Veo update is limited to Flow, Gemini API and Vertex AI), export.

Pricing & Access

Per Google’s plans page (one.google.com, fetched 2026-10-02), monthly Flow credits are: Google AI Plus 200, AI Pro 1,000, AI Ultra 10,000 or 25,000 (two Ultra tiers). Subscription prices differ by region and are not recorded here. Current prices: see https://one.google.com/about/google-ai-plans/

• Per the Flow help page: Omni at 360p costs half the credits of 720p, and Omni 360p can be upscaled to 720p free for Pro and Ultra; the page gives no per-model credit table.
• Gemini Omni requires a paid plan; Nano Banana image generation in Flow is free of credits for free users (Flow February 2026 update; free users also get daily credits, amount not stated by Google in the pages fetched).

Safety & watermarking

Outputs carry an invisible SynthID watermark (Omni clips included). Visible-watermark and content-policy rules per plan are not covered here; see Google’s Flow help pages.

Comparison with alternatives

Tool	Notes
Google Flow	Unified image + video + editing, bundled into Google AI plans
Runway	Dedicated video platform (see runway)
Sora	App/web ended 2026-04-26 per vault audit; API sunset scheduled (see openai-sora)
Midjourney	Image/video (see midjourney)

Limitations

Credit-metered generation; Omni clips limited to 10 s and some Omni voice features are region-limited; Veo 3.1 remains “preview” in the API.

Resources

• Flow: labs.google/fx/tools/flow
• Flow models help page: https://support.google.com/flow/answer/16352836
• Flow updates (Feb 2026): https://blog.google/innovation-and-ai/models-and-research/google-labs/flow-updates-february-2026/

Related

google-veo, nano-banana-pro, nano-banana-2, gemini-3-x-family, Google DeepMind, SynthID (watermarking technology)

Sources

Accessed 2026-10-01.

• https://support.google.com/flow/answer/16352836 (models in Flow: Veo 3.1 variants, Gemini Omni Flash 1.1, Nano Banana Pro/2/2 Lite)
• https://support.google.com/flow/answer/16353334 (Flow Agent, frames, references)
• https://blog.google/innovation-and-ai/models-and-research/google-labs/flow-updates-february-2026/
• https://blog.google/innovation-and-ai/technology/ai/veo-3-1-ingredients-to-video/ (2026-01-13)
• https://ai.google.dev/gemini-api/docs/models (Nano Banana 2/Lite/Pro stable; Veo 3.1 preview; Imagen 4 shut down)
• https://gemini.google/subscriptions/ (Plus/Pro/Ultra plans and Flow credits)
• https://www.redsharknews.com/google-io-2026-biggest-announcements-creatives (Gemini Omni in Flow, Ultra plan change; secondary)
• https://costgoat.com/pricing/google-flow (credits, free daily credits; secondary)
• https://aibusiness.vc/tools/google-io-2026-gemini-35-flash-price-war and https://www.cloudzero.com/blog/gemini-pricing/ (Ultra tiers; secondary)
• https://one.google.com/about/google-ai-plans/ (credits per plan; German-locale summary; accessed 2026-10-02)
• https://support.google.com/flow/answer/16352836 and https://ai.google.dev/gemini-api/docs/models (re-fetched 2026-10-02: Veo 3.1 Lite/Fast/Quality, Gemini Omni Flash 1.1, Nano Banana Pro/2/2 Lite; Imagen 4 shut down)