Google DeepMind’s current flagship text-to-image model. Served through four surfaces: the consumer Gemini app, Google AI Studio for free testing, the Gemini API for developers, and Vertex AI for enterprise production.
The model ships in three tiers: Fast, Standard, and Ultra. API pricing runs $0.02, $0.04, and $0.06 per image. Every output carries a SynthID watermark for provenance.
System Verdict
Pick Imagen 4 if your workflow lives inside Google’s stack. Native generation inside Docs, Slides, and Gmail removes the copy-paste step that breaks momentum in Workspace-heavy teams. Developers get predictable per-image pricing through the Gemini API with no subscription required. Vertex AI covers enterprise with SLAs, SSO, and adjustable safety filters.
Skip it if you need aesthetic ceiling. Midjourney still beats Imagen 4 on cinematic and stylized output. Flux beats it on open weights and LoRA fine-tuning. Ideogram beats it on dense text-in-image reliability for logos and posters.
Who pays which tier: Google AI Studio free for experimentation, Gemini API pay-per-image for low-volume apps, Google AI Pro $19.99/mo for individuals using Gemini + Imagen 4 in Workspace, Google AI Ultra from $100/mo or $200/mo for heavier Google media and agent workflows, Vertex AI for production with compliance needs.
Key Facts
| Model family | Imagen 4 Fast · Imagen 4 Standard · Imagen 4 Ultra |
| API pricing | Fast $0.02 · Standard $0.04 · Ultra $0.06 per image |
| Resolutions | Up to 2K standard · 4K on Ultra via token-metered output |
| Free testing | Google AI Studio playground (capped daily) |
| Consumer bundle | Google AI Pro $19.99/mo · Google AI Ultra from $100/mo, with $200/mo top tier |
| Enterprise | Vertex AI with Google Cloud billing, SSO, audit |
| Watermark | SynthID embedded in every output |
| Text rendering | Materially improved over Imagen 3 for signage and typography |
| Open weights | None |
| Best-in-class for | Workspace-embedded generation, per-image API billing |
Every model and API-pricing data point was last verified against vendor documentation on 2026-05-13. Google AI subscription pricing was refreshed on 2026-05-20 after Google I/O. Model IDs on the API surface are imagen-4.0-fast-generate-001, imagen-4.0-generate-001, and imagen-4.0-ultra-generate-001. See Sources.
What it actually is
One image model served through four entry points. Consumer users hit it through the Gemini app and get Imagen 4 generations bundled with Google AI Pro ($19.99/mo) or the current Google AI Ultra tiers that now start at $100/mo.
Developers access the same model through ai.google.dev) at per-image rates, or through AI Studio’s browser playground for free testing. Enterprise runs through Vertex AI, where Imagen 4 usage appears as line items on the Google Cloud bill alongside inference, storage, and logging.
The moat is integration depth. No other image model ships native inside Docs, Slides, Gmail, and Sheets. The Ultra tier also leans into 4K output, which matters for print and large-format signage that Midjourney and Flux still require an upscale pass to reach.
When to pick Imagen 4
- You work in Google Workspace every day. Generating inside Slides without export-reimport is the killer feature.
- You need predictable per-image API billing. Fast at $0.02 is the cheapest tier among major closed models for prototype apps.
- You need in-image text for signage, packaging, or vintage-style ads. Imagen 4 materially improved over Imagen 3 on legibility.
- You need SynthID provenance. Every output carries an embedded watermark, which matters for news and regulated publishing.
- You need Vertex AI for production. Google Cloud compliance, SLAs, and adjustable safety filters beat consumer endpoints for enterprise workloads.
When to pick something else
- Maximum aesthetic quality for cinematic, stylized, or illustrative work: Midjourney. Still the reference point for visual polish.
- Open weights, LoRA fine-tuning, or self-hosting: Flux. Imagen 4 offers none of those.
- Dense text rendering for logos and posters: Ideogram remains the safer pick for text-critical production.
- Image generation bundled with a chat assistant you already pay for: GPT Image 2 inside ChatGPT.
- Video generation alongside images: Gemini Omni for Google-native conversational video edits, Runway for production control, or Veo 3.1 when published Google API pricing matters.
Pricing
Consumer subscription pricing via gemini.google/subscriptions:
| Plan | Price | What you get |
|---|---|---|
| Gemini (free) | $0 | Limited daily Imagen 4 generations in the Gemini app |
| Google AI Pro | $19.99/mo | Gemini 3.1 Pro, Imagen 4 at consumer limits, 2TB storage |
| Google AI Ultra | From $100/mo; top tier $200/mo | Higher Gemini, media, agent, and Antigravity limits; exact Imagen limits should be checked in-account |
API and enterprise pricing via ai.google.dev/gemini-api/docs/models/imagen and cloud.google.com/vertex-ai/generative-ai/pricing:
| Tier | Per-image | Typical use |
|---|---|---|
| Imagen 4 Fast | $0.02 | Prototype apps, high volume at lower fidelity |
| Imagen 4 Standard | $0.04 | Production default |
| Imagen 4 Ultra | $0.06 | 4K output, campaign hero images |
| Google AI Studio | $0 | Browser playground for free testing (capped) |
| Vertex AI | Per-image + Cloud billing | Enterprise production with SLAs |
API prices verified 2026-05-13 via Gemini API docs and Vertex AI pricing. Google AI subscription prices refreshed 2026-05-20 via Google AI subscriptions and the May 19 I/O subscription update. Each Gemini API request accepts up to a 480-token prompt and returns 1 to 4 images.
Against the alternatives
| Imagen 4 | Midjourney V7 | GPT Image 2 | |
|---|---|---|---|
| Aesthetic ceiling | Strong photoreal, mid on stylized | Strongest stylized/cinematic | Strong, softer look |
| Text rendering | Materially improved, not Ideogram-grade | Improved on V7 | Good |
| API access | Full public API, per-image | None public | OpenAI API |
| Workspace integration | Native in Docs, Slides, Gmail | None | Via ChatGPT |
| Open weights | No | No | No |
| Watermark / provenance | SynthID on every output | None | C2PA metadata |
| Best viewed as | Google-stack integrator | Visual-quality specialist | Bundled generalist |
Failure modes
- Aesthetic ceiling trails Midjourney. Photoreal is strong, but cinematic composition and stylized illustration still look safer and more corporate out of the box.
- Safety filters reject more than competitors. Nudity, public figures, violent imagery, and some medical or historical reference get blocked where Flux or Stable Diffusion self-hosted would pass.
- Rate limits on consumer tiers. Even Google AI Pro caps daily Imagen 4 generations; hitting the wall during a deadline is a known failure mode.
- Vertex AI billing is opaque. Charges land inside the broader Google Cloud invoice alongside inference, storage, and logging. Cost attribution for a single campaign takes real bookkeeping.
- No open weights. Every workflow that needs offline, air-gapped, or fine-tuned generation has to use Flux or Stable Diffusion instead.
- Free AI Studio testing has undocumented daily caps. The playground is free but not guaranteed. Production apps should move to metered API.
- Consumer and API outputs both carry SynthID. Fine for most uses, but workflows that require clean pixel output without embedded watermarks need a different model.
Methodology
This page was produced by the aipedia.wiki editorial pipeline, an automated system that ingests vendor documentation, verifies pricing and model details against primary sources, and generates the editorial analysis you are reading. No individual human wrote this review. Scoring follows the four-dimension rubric at /about/scoring/ (Utility × Value × Moat × Longevity, unweighted average). Last verified 2026-05-20 against the May 19 Google I/O subscription update, the Imagen 4 Gemini API launch post, Gemini API Imagen docs, Vertex AI generative pricing, Google AI subscriptions, and the SynthID overview.
FAQ
Is Imagen 4 free to use? Partly. Google AI Studio allows free testing in the browser playground at capped daily volumes. The Gemini app on the free tier also includes limited Imagen 4 generations. Unlimited consumer access requires Google AI Pro at $19.99/mo.
How much does Imagen 4 cost on the API? Fast is $0.02 per image, Standard $0.04, Ultra $0.06, verified on ai.google.dev. Vertex AI uses the same per-image rates integrated into the Google Cloud bill.
Can Imagen 4 images be used commercially? Yes. Google permits commercial use of outputs from both consumer tiers and Vertex AI, subject to the standard acceptable-use policy. SynthID watermarking remains embedded in the output.
Imagen 4 vs Midjourney V7 for quality? Midjourney still wins on stylized, cinematic, and illustrative work. Imagen 4 is competitive on photoreal, closer than Imagen 3 was, and wins outright on in-image text legibility for signage-style prompts. For portfolio-grade art, Midjourney remains the default.
What is SynthID? Google’s provenance watermarking system. Every Imagen 4 output carries an invisible watermark detectable by Google’s verification tools. Details on deepmind.google/technologies/synthid.
Is Imagen 4 available outside Google? No. The model is served only through Google surfaces: Gemini app, AI Studio, Gemini API, and Vertex AI. No open weights and no third-party licensing.
Sources
- Imagen 4 Gemini API launch: availability across Gemini API and AI Studio
- Gemini API Imagen docs: model tiers, per-image pricing
- Vertex AI generative pricing: enterprise per-image rates
- Google AI subscriptions: Google AI Pro and Ultra consumer plans
- SynthID overview: provenance watermarking
Related
- Category: AI Image Generation
- Google media tools: Gemini Omni · Google Veo 3.1 · Gemini
