What’s actually slowing this PC down?
Recommended Free Tools
Pick the symptom - the matching free tool is one click away.
ThatPainter is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Learn More
Google Imagen 4’s launch pricing was simple and competitive: $0.02 per image for Fast, $0.04 for Standard, and $0.06 for Ultra. But that is now historical information. Google’s documentation says the Imagen 4 API endpoints were deprecated and scheduled to shut down on August 17, 2026, so as of August 18, 2026, Imagen 4 should not be treated as a current production option.
The launch still matters because Google paired a three-tier image API with a much broader consumer subscription called Google AI Ultra. Those two products shared the word “Ultra,” but they were not the same thing—and paying for the $249.99-per-month launch subscription did not provide unlimited Imagen 4 API images.
#1 Best Overall
What Google actually launched
Imagen 4 was introduced on May 20, 2025, as Google’s next-generation text-to-image model. Google emphasized improved typography, photorealism, color and scene detail, prompt adherence, speed, and output of up to 2K resolution. The announcement covered Google’s consumer products as well as developer access through the Gemini ecosystem.
It was not one single launch event:
- May 20, 2025: Google introduced Imagen 4 during its I/O-era announcements. Google’s announcement described the model’s broader product integration.
- June 24, 2025: Imagen 4 and Imagen 4 Ultra entered paid preview in the Gemini API, with limited free testing in Google AI Studio. Google’s paid-preview announcement listed Standard at $0.04 and Ultra at $0.06 per output image.
- August 15, 2025: The Imagen 4 family became generally available in the Gemini API and AI Studio. Google added Imagen 4 Fast at $0.02 per image. The general-availability announcement also described up-to-2K output for Imagen 4 and Ultra.
- August 17, 2026: Google’s current documentation lists the shutdown date for all Imagen 4 endpoints.
That timeline separates the original product announcement, preview pricing, general availability, and eventual retirement. A launch article that collapses them into a single date gives an incomplete picture.
At different stages, Imagen 4 was available through the Gemini API and Google AI Studio, while consumer access appeared in Gemini and Google creative products. Enterprise users could also encounter Google Cloud or Vertex AI pathways, subject to the product and region. Consumer access and developer API access were separate experiences with separate commercial terms.
For the developer API, Google documented these model IDs:
imagen-4.0-generate-001imagen-4.0-ultra-generate-001imagen-4.0-fast-generate-001
The model documentation listed a 480-token text input limit and one to four output images per request. Those identifiers should now be treated as legacy rather than as a recommendation for a new integration.
The three Imagen 4 tiers
| Model | Launch/GA price | Positioning | Best historical use |
|---|---|---|---|
| Imagen 4 Fast | $0.02 per output image | Speed and volume | Rapid exploration and high-volume generation |
| Imagen 4 Standard | $0.04 per output image | General-purpose quality | Everyday marketing, design, and illustration work |
| Imagen 4 Ultra | $0.06 per output image | Higher alignment and detail | Prompts where adherence and finish mattered most |
Fast cost half as much as Standard and one-third as much as Ultra. At the headline generation level, 1,000 images would have cost approximately $20 with Fast, $40 with Standard, or $60 with Ultra.
Those are nominal generation costs, not the cost of finished assets. They exclude taxes, retries, moderation or orchestration services, storage, image processing, upscaling, and the engineering required to build a production workflow. If a team needs four attempts to obtain one usable image, its effective cost is several times the listed request price.
Google’s pricing page stated that these Imagen 4 models did not have a free API tier and used paid-tier, per-image billing. The simplicity was useful: developers could estimate generation costs without converting token counts into image budgets. It did not guarantee that the cheapest tier delivered the lowest cost per successful asset.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsDo not confuse Imagen 4 Ultra with Google AI Ultra
Imagen 4 Ultra was a model. Google AI Ultra was a subscription. They were separate products with separate billing systems.
Imagen 4 Ultra was the premium member of the Imagen 4 API family. It cost $0.06 per generated image during the published preview and general-availability pricing period.
Google AI Ultra was a Google One subscription introduced in the United States at $249.99 per month in May 2025. Google described it as a bundle offering higher Gemini usage limits and access to services including Flow, its AI filmmaking tool using Veo, Imagen, and Gemini, plus Whisk benefits, Google app integrations, storage, and other Google One features. See Google’s launch announcement.
Google later announced a $100-per-month AI Ultra tier and reduced the top tier from $250 to $200 per month. The company also said the $200 plan would have a usage limit up to 20 times higher than the Pro plan, with parts of the Gemini ecosystem moving toward compute-based rather than simple daily limits. These changes came from Google’s 2026 subscription update, not from a change that turned the subscription into an Imagen 4 API allowance. Google’s subscription update is the relevant source for those later terms.
Recommended Free Tools
In practical terms, Google AI Ultra could be attractive to a creator who wanted a broad bundle of Gemini, Flow, Whisk, storage, and connected Google features. It was not a metered image API, and the available evidence does not support describing it as a subscription containing unlimited $0.06 Imagen 4 generations.
What Imagen 4 promised
Google’s product material highlighted several improvements:
- Typography: better rendering of words and lettering inside generated images.
- Photorealism: more detailed, realistic scenes and subjects.
- Prompt adherence: especially in the Ultra tier, according to Google.
- Resolution: up to 2K output for Imagen 4 and Imagen 4 Ultra at general availability.
- Speed: Google described Fast as up to 10 times faster than the previous model.
- Provenance: Google said generated images included SynthID invisible watermarking.
These are Google’s stated capabilities and performance claims, not proof that Imagen 4 universally outperformed every competing model. Google’s own Imagen overview acknowledges continuing weaknesses. Complicated compositions may contain artifacts, while small faces, thin structures, and text can still fail. Centered images and nonsensical prompts can also produce unreliable results.
SynthID can help identify AI-generated material in supported contexts, but watermarking does not by itself establish commercial safety, copyright clearance, or indemnity. Those questions require checking the applicable product terms and the way the image was created and used.
Free tools Windows power users keep installed
One-click scans. No signup required.
How the pricing compared with rivals
Google’s pricing looked aggressive in the context of premium image APIs, but “cheapest” is not a universal conclusion because competitors used different billing units and supported different workflows.
OpenAI GPT Image 1
OpenAI’s official documentation listed quality- and resolution-dependent prices for GPT Image 1, including:
- Low-quality 1024×1024: $0.011 per image
- Medium-quality 1024×1024: $0.042 per image
- High-quality 1024×1024: $0.167 per image
OpenAI can also bill image and text tokens separately in some workflows. That makes a direct comparison with Imagen 4’s flat per-output-image prices imperfect. A low-quality GPT Image 1 request was cheaper than Imagen 4 Fast, while its medium-quality figure was close to Imagen 4 Standard and its high-quality figure was substantially higher. GPT Image 1’s documented capabilities also include image input and editing workflows, so feature fit may matter more than the nominal generation price. The model page currently labels GPT Image 1 as deprecated as well; buyers should check its current status before adopting it. OpenAI’s documentation contains the detailed pricing and model information.
Adobe Firefly
Adobe Firefly competed on a different axis: it bundled generation and editing into a creative-production ecosystem rather than presenting only a simple image API. On its U.S. plans page, Adobe listed Firefly Standard at $9.99 per month and Firefly Pro at $19.99 per month. The page also showed Pro Plus at a regular $49.99 per month and Premium at a regular $199.99 per month, with monthly generative-credit allocations ranging from 2,000 to 50,000 depending on the plan. Temporary offers and regional terms can change.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Firefly also promotes access to multiple partner models. For Photoshop, Express, agency, and other Adobe-centered workflows, the value may come from editing tools, credit bundles, and application integration rather than from the lowest possible per-image API cost. See Adobe’s current plans page for live terms.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.The useful comparison is cost per usable asset
For a real production decision, compare more than the request price:
- Successful output rate: How often does a request produce an asset that needs no regeneration?
- Typography and layout: Important for posters, labels, thumbnails, and branded marketing graphics.
- Photorealism and anatomy: Relevant for people, products, editorial imagery, and advertising.
- Speed and queues: A cheap model can be expensive if it slows an interactive workflow or requires repeated attempts.
- Editing and references: Image-to-image, reference-image, masking, and iterative editing can reduce the need to regenerate from scratch.
- Resolution and aspect ratios: “Up to 2K” does not mean every interface, ratio, or workflow exposes identical output options.
- Commercial terms: Licensing, indemnity, training-data policies, and customer protections may outweigh a few cents per request.
- Operational durability: Rate limits, regional availability, API maturity, and deprecation policy affect the cost of maintaining an integration.
- Provenance: Watermarking and content credentials can matter for publishing and review workflows.
For example, a $0.02 Fast image that must be regenerated three times may be less economical than a $0.06 Ultra result that works on the first attempt. Conversely, a large thumbnail-generation service may reasonably favor Fast if small quality differences do not affect user outcomes.
The current problem: Imagen 4 has passed its shutdown date
As of August 18, 2026, the most important fact is availability, not launch pricing. Google’s Imagen model documentation says the Imagen 4 endpoints were deprecated and scheduled to shut down on August 17, 2026.
That creates a concrete failure mode for old tutorials and applications. A request using an imagen-4.0-*-generate-001 model ID may fail because the endpoint has been retired. Developers should not begin a new production integration against Imagen 4, and should not treat its former $0.02–$0.06 prices as an active purchasing option.
Google’s own migration guidance is currently inconsistent. The Imagen model page says to migrate to Gemini 3.1 Flash Image, while the Gemini API pricing page says to migrate to Gemini 2.5 Flash Image. Because those official references disagree, do not change a model ID based on an old article or choose between them blindly. Check Google’s live model and pricing documentation, confirm that the target is available to your account and region, and then test:
- Output quality, typography, composition, and aspect ratios.
- Input and output limits, latency, rate limits, and safety behavior.
- Per-image or token-based pricing and the cost of retries.
- Commercial-use terms, provenance features, and data handling.
- Fallback behavior and deprecation notices before deploying to production.
The available documentation does not establish that Imagen 4 remains a supported fallback after the stated shutdown date. A version-pinned legacy integration should therefore be treated as a migration liability, not as a dependable contingency.
What the launch meant
Imagen 4’s commercial strategy was significant at launch because it gave developers a clear ladder: pay $0.02 for speed, $0.04 for the general model, or $0.06 for the premium tier. Google then surrounded that API offering with consumer products and the Google AI Ultra subscription, making image generation part of a larger Gemini, Flow, Whisk, and Google Workspace story.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
That packaging did not eliminate the differences between a subscription and an API, nor did it prove that Google offered the best image quality or the lowest total cost. It offered a straightforward pricing signal and a broad ecosystem, while leaving buyers to evaluate output quality, retries, creative controls, licensing, and operational reliability.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
