iro (色)
Hanzo's image generation & editing model

Iro

Iro (iro, iro-1) is Hanzo's image generation and editing model, served from Hanzo's inference fabric at api.hanzo.ai under the canonical id iro (open foundation: zen3-image / zenlm/zen-artist). It renders one image per request from a text prompt at any aspect ratio from a 9:21 portrait to a 21:9 banner, and supports image editing on /v1/images/edits.

curl https://api.hanzo.ai/v1/images/generations \
  -H "Authorization: Bearer $HANZO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"iro","prompt":"a white canvas with a soft dot-grid, one dark rounded terminal showing a small diff, thin black contour lines flowing out of it, monochrome","size":"1792x1024"}'
{"data":[{"b64_json":"/9j/4AAQSkZJRgABAQAAAQABAAD…"}]}

That is the whole API: the OpenAI images schema, one model id (iro), no vendor vocabulary to learn. Iro answers b64_json whatever response_format asks for.

What it renders

Modes text → image (/v1/images/generations), image + prompt → image (/v1/images/edits)
Model IDs iro, iro-1 (open foundation: zen3-image, zen-image)
Output JPEG, base64 in the OpenAI data[].b64_json envelope
Images per request 1
Aspect ratios 9:21 5:11 1:2 7:13 3:5 2:3 3:4 6:7 1:1 7:6 4:3 3:2 5:3 13:7 2:1 11:5 21:9
Sizes any size you send; Iro fits it to the nearest ratio it can draw
Latency 1.6 s to 4.4 s per render, measured on production hardware

Sizes

You may ask for the size you actually want — 1792x1024, 1024x1792, 1536x1024, 256x256. Iro fits the request to the nearest renderable shape instead of refusing it, because each edge is on a 16-pixel grid between 512 and 1568, and the whole image is at most 1,062,400 pixels. 1792x1024 becomes 1344x768 — the same 7:4 ratio, the largest render that fits.

aspect_ratio wins when you send it, and drops size/width/height.

you ask you get ratio
1024x1024 1024x1024 1:1
1792x1024 1344x768 7:4
1024x1792 768x1344 3:7
1536x1024 1248x832 3:2
512x512 512x512 1:1
256x256 512x512 lifted to the floor

Creative AI Lineup

Each creative modality in Hanzo has its own dedicated model and repository:

Model Modality Endpoint Repository Open Foundation
Iro (iro) Image generation & editing POST /v1/images/generations hanzo-inc/iro zen3-image (zenlm/zen-artist)
Koma (koma) Video generation & motion POST /v1/videos/generations hanzo-inc/koma zen-director (zenlm/zen-director)
Koe (koe) Voice & dialogue synthesis POST /v1/audio/voice hanzo-inc/koe zen3-tts (zenlm/zen-dub)
Oto (oto) Foley & sound effects POST /v1/audio/foley hanzo-inc/oto zen-foley (zenlm/zen-foley)
Uta (uta) Music & score generation POST /v1/audio/music hanzo-inc/uta zen-musician (zenlm/zen-musician)
Mori (mori) 3D assets, meshes & scenes POST /v1/videos/generations hanzo-inc/mori zen-3d (zenlm/zen-3d)

Surfaces

Iro is served by hanzo-inc/zen and declared in hanzo/universe (charts/app/files/zen/catalog.yaml), with first-party Studio integration in hanzo-inc/studio (image.generate / image.edit).

License

The contents of this repository are licensed under Apache-2.0. The proprietary iro-1 weights are served through the Hanzo API under the Hanzo Terms; the open foundation zen3-image is available under Apache-2.0 at huggingface.co/zenlm/zen3-image.

S
Description
Iro: Hanzo's image model — text-to-image generation & editing at any aspect ratio
Readme
37 KiB
Languages
Markdown 100%