Guide
MiniMax H3 Image Developer Guide: Capabilities, API Parameters, and Pricing
Auf dieser Seite
What MiniMax H3 Image Offers
MiniMax H3 Image is an image family with two routes: text-to-image and image-to-image. Both return one image per request and share the same resolution tiers and per-image base prices. Route choice is about input shape, not about a different rate card.
Use text-to-image when you only have a prompt. Use image-to-image when the request must edit or combine existing pixels — that route requires image_urls as well as prompt.
API Parameters
Shared optional fields on both routes:
prompt(required, string, 1–5000 characters). On text-to-image, describe the image. On image-to-image, describe the edit and refer to inputs in order as Picture 1, Picture 2, and so on. Prompts are automatically expanded for more detail.resolution(optional, string, default2K). One of1K(about 1 MP),2K(about 4 MP), or4K(about 16 MP).aspect_ratio(optional, string). Enum:auto,1:1,5:4,4:5,4:3,3:4,3:2,2:3,16:9,9:16,21:9. Text-to-image defaults to1:1; image-to-image defaults toauto, which lets the model choose based on the prompt and reference images.output_format(optional, string, defaultjpeg).jpegorpng. PNG files are much larger (about 12 MB at 2K, 30 MB at 4K).seed(optional, integer, 0–2147483647). Set it for reproducible results; omit it for a random seed.
Image-to-image also requires image_urls: 1–16 images (JPG, PNG, or WEBP, each side 128–8192 px, up to 20 MB). Each reference image is billed as an additive surcharge, described below.
Official Per-Image Pricing
Both routes use the same official resolution-tier prices:
- 1K: $0.0129 per image
- 2K: $0.0129 per image
- 4K: $0.0393 per image
The default resolution is 2K, so a request that omits the field is billed at $0.0129, not at the 4K rate. 4K is the only higher tier.
On image-to-image, each reference image adds an official surcharge of $0.0036. The charge is additive, not a multiplier of the tier price:
total = resolution-tier price + (reference count × $0.0036)
Examples: a 2K image with no references is $0.0129; a 2K image with three references is $0.0129 + 3 × $0.0036 = $0.0237; a 4K image with two references is $0.0393 + 2 × $0.0036 = $0.0465. Text-to-image has no image_urls field, so it has no reference surcharge.
Confirm the live panels on the two route pages if you need to re-check, and integrate against these official figures.
Integration Checklist and Common Pitfalls
Before you ship, walk this list.
- Validate inputs client-side. Enforce the 1–5000 character prompt range, the 1–16 reference array bounds, the 128–8192 px per-side limits, and the per-file size cap before the request leaves your service. Server-side rejections are slower and harder to debug.
- Budget payload sizes. Decide up front whether you'll accept 30 MB PNG results at 4K and multi-reference request bodies, and whether your storage and CDN tiers can absorb them.
- Handle seed deliberately. If you need reproducible output for review, regression tests, or A/B comparison, set and persist
seed. If you want variation across calls, omit it — don't set a seed and then wonder why every result looks the same. - Choose the right route. Prompt-only work goes to text-to-image; anything referencing existing pixels goes to image-to-image. Remember the differing
aspect_ratiodefaults (1:1versusauto) so your first test isn't a surprise. - Count your references. Each reference adds $0.0036 on top of the resolution-tier price. Trim to the minimum set that accomplishes the edit.
- Read the route docs. The parameter schemas for text-to-image and image-to-image are the source of truth for enums and limits, and official prices are $0.0129 at 1K and 2K, $0.0393 at 4K, plus $0.0036 per reference image on image-to-image.
MiniMax H3 Image is a small, predictable surface: two routes, a handful of parameters, and a resolution-tiered per-image bill. Integrate it with input validation and deliberate resolution choices, and most of the usual surprises never appear.

Ähnliche Artikel
Alle Artikel ansehen →-
Guide
gpt-6.2-sol Getting Started Guide: Capability Boundaries, API Integration, and Token Pricing Essentials
Integration decisions and practical essentials for engineering teams
-
Guide
gpt-6.2-astra Getting Started Guide: Text Model Capability Boundaries, Integration Essentials, and Cost Control
-
Guide
Tripo H3.1 Getting Started Guide: Choosing Among Three Input Modes, Parameters, and Pricing
A practical guide for developers on 3D generation integration decisions and cost control