Uni-1.1
Uni-1.1 generates and edits images from text and visual references. Luma describes a decoder-only transformer that processes text and image tokens together. This supports instructions that combine composition, subject identity and changes to a supplied image.
The identifier and model version differ
The API name does not contain the .1 version suffix. Luma’s training disclosure identifies Uni-1.1 as an update to Uni-1 and Max as a higher-quality tier in the same family. It also discloses an unnamed third-party open-weight component used before Luma’s further training. This does not make Uni’s final weights a public download.Luma’s Uni training-content disclosure
Use references to specify the change
The API announcement describes edits such as replacing a background, changing lighting or applying a reference style. Text and reference images can be combined in one request, and output can be PNG or JPEG. The source image and the references have different roles: the first is the image being changed, while the others guide the result.Uni-1.1 API announcement
Practical guidance. For a product image, keep the original product as a reference and describe only the scene changes you need. Inspect labels and product proportions before publication. This review remains necessary even when a generated image looks plausible.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | API, Provider product |
| Model identifiers | uni-1uni-1-max |
| API identifier | uni-1 serves Uni-1.1; uni-1-max selects the higher-quality tier.Luma’s Uni training-content disclosure |
| Version history | Uni-1.1 replaced Uni-1 behind the same API identifier.Luma’s Uni training-content disclosure |
| References | Up to nine for generation; up to eight for editing because the source image uses one slot.Luma Agents API models |
| Output | 2K images. Text-to-image costs 14 credits on uni-1 or 34 on uni-1-max; account terms determine the dollar value of each credit.Luma Agents API prices and capacity plans |
Related models
A video model for generation, editing, extension and reframing from text, image anchors or existing video.
Launch date not confirmedPreviewAPI · Provider product