Provider
Google AI Models on Martini
Google ships three of the strongest models on Martini: Veo for cinematic video, Imagen 4 for photoreal image, and Nano Banana 2 for character-consistent edits. The trio covers brand visuals, ads, and multi-shot stories without leaving the canvas, and shares one credit wallet with every other provider.
About Google
Google's generative stack on Martini centers on three model families. Veo is the cinematic text-to-video and image-to-video model — it competes head-on with Sora 2 and Kling 3 for photoreal motion and is widely seen as the cleanest model for brand-safe footage. Imagen 4 is Google's flagship image model with strong photoreal lighting, accurate human faces, and balanced color science. Nano Banana 2 (the next-gen Imagen-backed editing model) is the workhorse for character consistency, reference-driven edits, and surgical inpainting where you need the same subject to appear shot after shot.
On Martini these three are wired into the canvas as drop-in nodes that interoperate with every other provider. A common pattern is Imagen 4 for the hero brand image, Nano Banana 2 for variants and character consistency across a campaign, then Veo for the motion pass — all running in one project file. Google's strengths show up most when you care about photographic realism, accurate skin and faces, or production-clean motion, which is why agency and product teams default to this stack for client work.
Pricing is metered per generation and shares your Martini wallet, so there's no separate Google AI Studio or Vertex AI bill to manage. Workspace billing keeps Google usage attributable per member, and if Imagen rejects a prompt for safety the same canvas can fall back to Flux, Midjourney, or Seedream without rebuilding the workflow. For teams already deep in Veo or Imagen, Martini is the way to keep using them while gaining a shared workspace, version history, and multi-provider redundancy.
Available Google models on Martini
Video models
Best use cases
- Cinematic brand video where photoreal motion and clean lighting matter more than stylized output
- Product visuals that need accurate skin tones, materials, and natural color science
- Character consistency across a campaign using Nano Banana 2 reference images
- Hero ad creative built in Imagen 4 then animated with Veo for the hero spot
- Multi-shot narrative video where each cut needs to look like the same world
- Brand visuals where typography is secondary to photographic realism
Recommended workflows
AI Character Consistency
Nano Banana 2 is the strongest reference-driven model for keeping a character locked across shots.
AI Image to Video
Imagen 4 produces a photoreal still that Veo can animate into a cinematic clip in one canvas.
AI Video Workflow
Veo, Imagen 4, and Nano Banana 2 cover the full image-to-video pipeline with brand-safe defaults.
AI Product Video Generator
Imagen 4 hero shots paired with Veo motion is the cleanest way to ship product video at brand quality.
Related how-to guides
Related features
Related reading
Other providers
Frequently asked questions
Do I need a Google Cloud or Vertex AI account to use Veo and Imagen on Martini?
No. Martini provides hosted access to Veo, Imagen 4, and Nano Banana 2 — you pay with Martini credits and never wire up Vertex AI or Google AI Studio.
How does Veo compare to Sora 2 and Kling 3?
Veo leads on photoreal brand-safe motion and natural color science. Sora 2 leads on physics-aware multi-shot narrative. Kling 3 leads on motion fidelity and character refs. Most teams use all three on the same canvas and pick per shot.
Is Nano Banana 2 better than Flux Kontext for character consistency?
Both are strong. Nano Banana 2 wins for photoreal humans and reference-image character lock. Flux Kontext wins for stylized illustration and structural edits. Run a quick A/B on Martini to pick.
Can I chain Imagen 4 into Veo for image-to-video?
Yes — connect an Imagen 4 image node to a Veo video node and Veo will animate the still into a cinematic clip with image-to-video mode.
Does Veo generate audio?
Veo supports audio generation alongside video for many models, so you can produce footage with ambient sound or dialogue without a separate voiceover step.
What if Imagen 4 rejects my prompt?
Google models apply safety filters. If a generation is blocked you can fall back to Flux, Midjourney, or Seedream from the same canvas without rebuilding the workflow.
Build with Google on Martini
Open the canvas and wire Google into the rest of your stack in minutes. Free to start — no card required.