GPT Image 2 on VM0. OpenAI's newer image generation model
OpenAI's newer image model for text-to-image generation and edits. Use it when you want the GPT Image workflow with stronger prompt adherence and flexible sizing.
Image / Text-to-image / Image edit
GPT Image 2 is OpenAI's newer image generation model on VM0. It keeps the GPT Image text-to-image and edit workflow while adding flexible sizing and stronger prompt adherence for production image tasks.
VM0 billing is tier-based, from around $0.007 per image at the low/standard tier to $0.481 at the high/large tier. The medium-standard tier (around $0.064 per 1024×1024 on VM0) is the sensible default for most built-in generation calls.
What is GPT Image 2?
April 2026 · OpenAI's newer GPT Image model, positioned above GPT Image 1 for prompt adherence and flexible output sizing.
GPT Image 2 is the newer OpenAI image model exposed through VM0's built-in image generation path. It supports text-to-image generation and image edits through the same agent-facing workflow as GPT Image 1.
The model is a good default when the caller wants OpenAI-style prompt following but needs more flexible output sizing than the standard GPT Image 1 tiers. It is priced by quality and size, so agents can keep draft loops cheaper and reserve high-quality output for final renders.
What's notable about GPT Image 2
Headline architecture and capability features.
Text-to-image and image-edit model exposed through OpenAI's GPT Image 2 endpoint on fal. Supports quality tiers and flexible image sizes; transparent backgrounds are not supported on VM0 for this model.
Specs at a glance
GPT Image 2 pricing
Vendor list price per generated unit.
How GPT Image 2 behaves in practice
Observed behaviour from production agent runs.
Prompt adherence
The main reason to pick GPT Image 2 over GPT Image 1 is stricter execution of detailed prompts, especially when the brief includes composition, aspect ratio and style constraints.
Flexible sizing
Supports VM0's flexible image-size path, so agents can request square, portrait, landscape and larger delivery sizes without switching model families.
Edit flows
Supports image-to-image edits and masks. Use it when a workflow needs to preserve a source image while changing a described region.
Cost
More expensive than GPT Image 1 at medium and high tiers. Keep draft loops on lower quality and reserve high quality for final output.
Best agent tasks for GPT Image 2
The agent that needs exact prompt following
When the prompt has specific framing, style or size requirements, GPT Image 2 is the safer OpenAI-image choice than GPT Image 1.
The image workflow that needs multiple aspect ratios
Marketing, social and website assets often need square, portrait and landscape outputs. GPT Image 2 fits those workflows without changing the prompt style.
The OpenAI-stack edit loop
If an agent is already using OpenAI models for planning and copy, GPT Image 2 keeps image generation in the same prompt-following family.
When to skip GPT Image 2
Skip GPT Image 2 when cost dominates or when true transparent-background output is required. GPT Image 1, GPT Image 1.5 or SeedDream 4 may be better depending on the constraint.
GPT Image 2 vs other models
GPT Image 2 vs GPT Image 1
GPT Image 2 is the newer, more flexible model with stronger prompt adherence. GPT Image 1 remains cheaper at common medium-standard settings and is still strong for stylised illustration.
GPT Image 2 vs SeedDream 4
SeedDream 4 is cheaper and photoreal-leaning. GPT Image 2 is the better OpenAI-stack choice when prompt adherence, edit workflow and flexible sizing matter more than lowest cost.
Bottom line: should you use GPT Image 2?
Pick GPT Image 2 when you want OpenAI image generation with stronger prompt adherence and flexible sizing. Use GPT Image 1 or SeedDream 4 when cost is the dominant constraint.
Frequently asked questions
How is GPT Image 2 priced?
It is priced by quality and size. On VM0, the medium-standard tier is about $0.064 per image and the high/large tier is about $0.481 per image.
Does GPT Image 2 support image editing?
Yes. VM0 exposes image-to-image editing and mask support for GPT Image 2.
Does GPT Image 2 support transparent backgrounds?
No. VM0's GPT Image 2 path does not support transparent-background output.
When should I use GPT Image 2 instead of GPT Image 1?
Use GPT Image 2 for stricter prompt adherence and flexible sizing. Use GPT Image 1 when the brief is simpler or cost matters more.
Alternatives
Using GPT Image 2 on VM0
Using GPT Image 2 on VM0
VM0 agents can call GPT Image 2 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.