GPT Image 1 on VM0. OpenAI's text-to-image model
OpenAI's text-to-image model with strong stylised illustration and editing. The natural pick when you want OpenAI's aesthetic and prompt-following style.
Image / Text-to-image / Image edit
GPT Image 1 is OpenAI's text-to-image model — the one most teams know as the model behind ChatGPT's image generation. Its strengths are stylised illustration, character work and image editing with text-driven masking, with prompt-following style that maps closely to what OpenAI's text models expect.
List price is tier-based, from around $0.011 per image at the low/standard tier to $0.25 at the high/large tier. The medium-standard tier (around $0.05 per 1024×1024) is the sensible default for most agent workloads.
What is GPT Image 1?
April 2026 · OpenAI's primary text-to-image model. Tier-priced across resolution and quality settings.
GPT Image 1 is OpenAI's production text-to-image model. It pairs natively with OpenAI's text models, so when an agent already runs on GPT-5.4 or GPT-5.5 the prompt style transfers cleanly and the edit-loop flow stays inside the OpenAI surface.
The model's stylistic strengths sit on illustration, character work and edits that preserve the original composition while changing a specific element. Photoreal output is solid but skews towards the OpenAI house style; teams that want a different aesthetic ceiling often reach for Flux Pro 1.1 Ultra or SeedDream 4 alongside it.
What's notable about GPT Image 1
Headline architecture and capability features.
Diffusion-based text-to-image with native edit support. Tier pricing scales by output resolution (standard / large) and quality (low / medium / high), with the medium/standard tier as the typical default. Inputs accept text plus optional reference images for edits and masks.
Specs at a glance
GPT Image 1 pricing
Vendor list price per generated unit.
How GPT Image 1 behaves in practice
Observed behaviour from production agent runs.
Stylised illustration
One of the strongest models for non-photoreal output — illustration, comic-style, painterly. Good fit when the deliverable is an illustration rather than a photo.
Edit flows
Native support for masked edits and text-driven local changes. Useful when an agent has to iterate on a specific region of an image rather than re-generating the whole thing.
Prompt style
Maps closely to OpenAI's text models' expectations. When the calling agent is already on GPT-5.4 or GPT-5.5, the prompts written by that agent transfer with little adjustment.
Cost
Tier-based — the medium/standard tier (~$0.05 per 1024×1024) is the typical default. The high/large tier hits $0.25 and is worth it only for delivery-grade output.
Best agent tasks for GPT Image 1
The illustration agent that delivers comic or hand-drawn style
Stylised output is where GPT Image 1 carries a real edge. Comic panels, painterly illustrations, hand-drawn-look icons — all land more reliably here than on the photoreal-leaning alternatives.
The edit-loop agent on the OpenAI stack
If the orchestrating agent is already on GPT-5.4 or GPT-5.5, keeping image generation inside the OpenAI surface (GPT Image 1) means the prompt style, edit semantics and structured outputs stay consistent across the run.
When to skip GPT Image 1
Skip GPT Image 1 when the deliverable is specifically photoreal (SeedDream 4 carries a higher photoreal ceiling).
GPT Image 1 vs other models
GPT Image 1 vs SeedDream 4
SeedDream 4 leads on photoreal aesthetic at a slightly lower price; GPT Image 1 leads on stylised illustration and edit flows.
GPT Image 1 vs Flux Pro 1.1 Ultra
Flux Pro 1.1 Ultra carries the highest aesthetic ceiling for hero-shot deliverables; GPT Image 1 is the natural OpenAI-stack default for everything else.
Bottom line: should you use GPT Image 1?
Pick GPT Image 1 when your agent is already on the OpenAI stack and you want stylised illustration or native edit flows. Escalate to Flux Pro 1.1 Ultra for photoreal hero shots; drop to SeedDream 4 when cost dominates.
Frequently asked questions
How is GPT Image 1 priced?
Tier-based — combinations of size (standard / large) and quality (low / medium / high). The medium/standard tier at ~$0.05 per 1024×1024 image is the typical default.
Does GPT Image 1 support image editing?
Yes. It accepts a reference image plus an optional mask and supports text-driven local edits as well as outpainting.
Can GPT Image 1 render text inside images?
Yes — short text strings render reliably; long text passages still struggle as with most diffusion models.
Is it multimodal as input?
Image-to-image edits accept reference images. The model is image-output only.
Alternatives
Using GPT Image 1 on VM0
Using GPT Image 1 on VM0
VM0 agents can call GPT Image 1 as part of an agent run, billed against your VM0 credits. The list price above is what the upstream provider charges; VM0 passes that through with the standard credit conversion.
Available on VM0 since April 2026.