Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.kymaapi.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

imagen-4 is Google’s default Imagen 4 tier — the balanced choice between latency and fidelity. Photoreal humans, sharp text rendering, rich compositional reasoning. Recommended starting point for hero/brand assets. Lives in the Imagen 4 family alongside imagen-4-fast (drafts) and imagen-4-ultra (print-ready). Same style language; pick by quality + cost.

​
Specs

FieldValue
Model IDimagen-4
CreatorGoogle
BackendVertex AI (imagen-4.0-generate-001, us-central1)
Best forPhotoreal portraits, branded scenes, hero images
Aspect ratios1:1 (default), 3:4, 4:3, 16:9, 9:16
Sample count n1–4
Pricing modeFlat per image
Default latency~10–20s
OutputBlob-hosted URL (Vercel CDN)

​
Pricing

VariantProvider costKyma list
imagen-4$0.040$0.054
Live source: GET https://kymaapi.com/v1/pricing.

​
Compared to other image models on Kyma

Strengthimagen-4imagen-4-fastimagen-4-ultraflux-2-progpt-image-2
Default photoreal★★★★★★★★★★★★★★★★★★★★★★
Speed★★★★★★★★★★★★★★★★★★★
Cost★★★★★★★★★★★★★★★★★★★
Text in image★★★★★★★★★★★★★★★★★★
Print 4MP+★★★★★★★★★★★★★★★★★

​
Use this when

  • You want a strong default photoreal output without paying premium pricing.
  • The brief is portraits, branded scenes, hero/marketing images.
  • Iteration is faster than imagen-4-ultra but fidelity beats imagen-4-fast.

​
Pick something else when

  • You need print-ready at 4MP+ → imagen-4-ultra.
  • You need legible text in the image → gpt-image-2 high tier.
  • You need multi-reference blending (up to 10 source images) → flux-2-pro.
  • You need to edit an existing image → nano-banana.

​
Example

curl -X POST https://kymaapi.com/v1/images/generations \
  -H "Authorization: Bearer $KYMA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "imagen-4",
    "prompt": "A photoreal golden retriever puppy in a field of sunflowers at sunset",
    "size": "1024x1024",
    "n": 1
  }'
The endpoint is async — poll GET /v1/jobs/{id} until status is succeeded.