model guide
GPT Image 2: AI Image Model Guide
GPT Image 2 guide with image generation modes, output tiers, Kansova credit costs, prompts, limitations, and alternatives.
TL;DR
An image model for detailed prompts, reference edits, and polished visual concepts.
Who this is for
- Marketing teams writing detailed art-direction prompts.
- Product designers iterating compositions and material treatments.
- Creators who need generation and editing in the same workflow.
Best for / Not best for
Best for
- Detailed instruction following
- Product compositions
- Image editing
- High-resolution concept output
Not best for
- Exact text reproduction
- Deterministic CAD output
- Long-form video
- Unlimited free batch generation
Parameters
| Parameter | Supported value | Notes |
|---|---|---|
| Output size | 1K, 2K, or 4K | 4K uses a higher credit tier. |
| Input | Text or reference image | Supports generation and editing workflows. |
| Aspect ratio | 1:1 through 21:9 | Available ratios depend on the selected provider model. |
| Outputs | One image per task | Run additional tasks for more variants. |
| Processing | Asynchronous | Kansova polls task status until completion. |
Pricing
Kansova bills generations in credits. Pro is $19/month for 500 credits; Studio is $39/month for 1,100 credits.
| Generation | Kansova credits | Cost methodology |
|---|---|---|
| 1K image | 1 credit | Provider rate varies by routed model |
| 2K image | 1 credit | Provider rate varies by routed model |
| 4K image | 2 credits | Provider rate varies by routed model |
Comparison
| Dimension | GPT Image 2 | Nano Banana | Midjourney | Stable Diffusion |
|---|---|---|---|---|
| Primary strength | Detailed prompts | Fast edits | Stylized imagery | Local control |
| Kansova access | Available | Available | Not currently public | Not currently public |
| Reference editing | Yes | Yes | Limited workflow | Depends on setup |
| Kansova cost | 1-2 credits | 1-2 credits | Not priced | Not priced |
| Best workflow | Directed concepts | Rapid variations | Art direction | Technical customization |
Prompt examples
- 01
Premium glass serum bottle on a pale blue plinth, soft water caustics, clean ecommerce lighting, 1:1. - 02
Replace the chair in this interior with a low sculptural stone bench while preserving camera angle and lighting. - 03
Minimal editorial still life of brushed steel objects on dark glass, precise reflections, restrained luxury campaign. - 04
Modern concrete courtyard house at dusk, warm interior light, native planting, realistic architectural visualization. - 05
Create three colorway concepts for this packaging while preserving its proportions and label hierarchy.
Known limitations
- Small typography and brand marks still need manual verification.
- Each request returns one image in the current routed workflow.
- Complex edits can change details that were not explicitly protected in the prompt.
- Provider queue and model availability affect completion time.
Alternatives
Frequently asked questions
How many credits does GPT Image 2 use?
Kansova currently charges 1 credit for 1K or 2K and 2 credits for 4K.
Does GPT Image 2 support editing?
Yes. The routed API supports text-to-image, image-to-image, and image editing.
How many images are returned?
The current workflow returns one image per generation task.
Last updated
2026-07-05
Sources and methodology
Capabilities were checked against the provider API material supplied to Kansova and the live Kansova controls on 2026-07-05. Credit figures are taken from the production pricing function; provider rates are reference figures and may change independently.
- EvoLink API documentation index
Provider documentation used to verify image modes and output tiers.
- Kansova product facts
Canonical list of supported workflows and current limitations.
- Kansova pricing
Current subscription and monthly credit allowances.