We piped product photography through this model and edit cycles dropped from days to minutes.
The GPT Image 2 API edits images (OpenAI GPT Image 2) from natural-language prompts. Optional masks for surgical inpainting, transparent-background output, three quality tiers (Low / Medium / High) across sizes from 1024×1024 up to 4K (3840×2160) — all through one REST endpoint, priced 5% below OpenAI's published rate. Representative tiers below; all 18 sizes plus `auto` are available via the API.
Input
The text prompt sent to the model
Output
SampleRun the model to replace this previewGPT Image 2 API: Image Editing
Use the GPT Image 2 API to edit images from natural-language prompts.
GPT Image 2 API edits images (OpenAI GPT Image 2) from natural-language prompts. Three quality tiers (Low / Medium / High), 1024–1536 px output, optional masks for inpainting, and transparent-background support — all through one REST endpoint.
Pricing on the GPT Image 2 API is 5% below OpenAI's published rate, held flat across quality tiers. The GPT Image 2 API uses one RouterBase key — no provider-specific SDKs.
Six reasons teams ship on the GPT Image 2 API
From mask-driven inpainting to 5% off official pricing — what the GPT Image 2 API gets right.
Natural-language editing
Skip masks and layers. Describe the change and the GPT Image 2 API renders it.
Mask-driven inpainting
Pass an optional mask for surgical edits — the model touches only the masked region.
Transparent backgrounds
PNG with alpha channel. Drop ecommerce assets onto any backdrop in one call.
5% off official
Pay 5% less than OpenAI's published rate; the GPT Image 2 API discount is flat across tiers.
REST + JSON
One endpoint, one key, no SDK lock-in. Plain JSON in, signed URL out.
Three quality tiers
Low / Medium / High render quality at 1024×1024 to 1024–1536 px, set per-call.
Quickstart
Get started with the GPT Image 2 API in 3 steps
From sign-up to your first GPT Image 2 API call in under 5 minutes.
Create a RouterBase API key
Sign up and generate sk-rb-… — one key works for the GPT Image 2 API and 200+ other models.
POST to the GPT Image 2 API endpoint
Send a base image, prompt, and optional mask in JSON. The GPT Image 2 API returns a signed URL within seconds.
Inspect the result
Each GPT Image 2 API response includes generation metadata for auditing and a render URL ready to embed.
Pay only for what you use
RouterBase passes through partner-tier pricing. Compared against the model's official published API rate.
| Size | Low Quality | Medium Quality | High Quality |
|---|---|---|---|
| 1024 × 1024 Square 1:1 | $0.0056 | $0.050 | $0.200 |
| 1024 × 1536 Portrait 2:3 | $0.0045 | $0.039 | $0.156 |
| 1536 × 1024 Landscape 3:2 | $0.0045 | $0.039 | $0.156 |
| 1024 × 2048 Tall 1:2 | $0.0038 | $0.034 | $0.135 |
| 2048 × 1024 Wide 2:1 | $0.0038 | $0.034 | $0.135 |
| 688 × 2048 Tall 1:3 | $0.0020 | $0.019 | $0.075 |
| 2048 × 688 Wide 3:1 | $0.0020 | $0.019 | $0.075 |
| 880 × 2048 Ultra-tall 9:21 | $0.0030 | $0.027 | $0.107 |
| 2048 × 880 Ultrawide 21:9 | $0.0030 | $0.027 | $0.107 |
| 1152 × 2048 Portrait 9:16 | $0.0045 | $0.040 | $0.161 |
| 2048 × 1152 Landscape 16:9 | $0.0045 | $0.040 | $0.161 |
| 1360 × 2048 Portrait 2:3 | $0.0060 | $0.052 | $0.209 |
| 2048 × 1360 Landscape 3:2 | $0.0060 | $0.052 | $0.209 |
| 1536 × 2048 Portrait 3:4 | $0.0070 | $0.063 | $0.253 |
| 2048 × 1536 Landscape 4:3 | $0.0070 | $0.063 | $0.253 |
| 2048 × 2048 Square 2K | $0.011 | $0.102 | $0.407 |
| 2160 × 3840 Portrait 4K | $0.011 | $0.095 | $0.340 |
| 3840 × 2160 Landscape 4K | $0.011 | $0.095 | $0.340 |
Your request will cost $0.050 per run (1024 × 1024 / Medium).With $1 you can run this model approximately 20 times.
Pay only for what you use. GPT Image 2 bills on OpenAI's token model — 5% below published rates: image output priced by size × quality (each of n images billed separately), plus your prompt text ($5 / 1M tokens) and, on edit, reference images ($8 / 1M). The table shows per-image output rates; Official = OpenAI's published rate.
What teams build with this image-edit model
Real production loads running on the GPT Image 2 API and the RouterBase model catalog.
We swapped three providers through RouterBase in one line of code when OpenAI had its outage. Our users never noticed.
Locked product detail across a 500-shot catalog with the GPT Image 2 API — masks plus prompt, no manual cleanup.
The GPT Image 2 API replaced three artist-hours of mask cleanup per shot. Our pipeline barely flinched.
Plugged the GPT Image 2 API into my existing OpenAI SDK with one base URL swap — less work than writing this testimonial.
RouterBase puts 200+ models behind one key. Our research team stopped asking for new provider accounts.
Background-removal moved to the GPT Image 2 API — transparent backgrounds and 1536 px renders behind one REST call.
We migrated 42 services to RouterBase over a weekend. The only PR comment we got was 'wait, that's all?'.
Pricing on the GPT Image 2 API is 5% below OpenAI's published rate — at our scale that's six figures saved per quarter.
Frequently Asked Questions
Common questions about the GPT Image 2 API.
The GPT Image 2 API is RouterBase's pass-through to OpenAI GPT Image 2 (image edit). It edits images from natural-language prompts with optional masks, transparent backgrounds, three quality tiers, and 1024–1536 px output.