ZMIME
Image model comparison · 2 models · Updated Nov 21, 2026

P-Image-OSS vs MiniMax H3 Image

Newest: MiniMax H3 Image · Most API hosts: MiniMax H3 Image. Not enough arena votes yet for a quality score, so this is a spec comparison.

  1. Pruna

    P-Image-OSS

    Released Nov 3, 2026

    • Text to image
    • Image editing
    • Arena—
    • From—
    • Max size—
  2. MiniMax

    MiniMax H3 Image

    Released Nov 10, 2026

    • Text to image
    • Image editing
    • Arena1,358
    • From—
    • Max size—
  3. Add a model

    Make it a three-way comparison.

01 — At a glance

Where they differ

Quality is the arena score from blind, side-by-side human votes. Price, length and resolution come from each lab’s own listings.

  • NewestMiniMax H3 ImageP-Image-OSS Nov 3, 2026 · MiniMax H3 Image Nov 10, 2026
  • Widest inputsSame inputsP-Image-OSS: Text prompts, Reference images · MiniMax H3 Image: Text prompts, Reference images
  • Image editingBoth edit imagesTakes a reference image as input: P-Image-OSS yes · MiniMax H3 Image yes
  • Self-hostingAll open weightsDownloadable weights (MiniMax H3 Image Community License)
  • Most API hostsMiniMax H3 ImageP-Image-OSS: 0 · MiniMax H3 Image: 1
02 — Side by side

Every spec in one table

Dashes mean the data is not published.

P-Image-OSS vs MiniMax H3 Image specifications side by side
SpecificationP-Image-OSSPrunaMiniMax H3 ImageMiniMax
TaskImage generation and editingImage generation and editing
Published capabilitiesAdapts the MiniMax H3 backbone to still-image generation and multi-reference edits.One checkpoint handles both text-to-image generation and editing.
AccessPublic weights, ComfyUI workflow and Hugging Face demoDownloadable BF16 diffusion transformer and inference resources
Operating limits2–16 megapixels and up to nine input images. The smallest published build is 20 GB.The publisher reports a working range of roughly 4–20 megapixels with best results around 16 MP.
SourcesPublisher documentation · Model weightsModel weights · Additional documentation
Quality (blind human votes)
Text to image arenaElo-style score · higher is better—1,198 ±11#24 of 86 · 3,285 votes
Image editing arenaElo-style score · higher is better—1,358 ±9#24 of 64 · 5,134 votes
Output
ProducesImagesImages
Inputs
Text promptsYesYes
Reference images (editing)YesYes
Access
WeightsOpenOpenMiniMax H3 Image Community License
API hosts—1
ReleasedNov 3, 2026Nov 10, 2026
03 — Questions

Which should you choose?

Which is better, P-Image-OSS or MiniMax H3 Image?

Neither model has enough arena votes for a reliable quality score yet, so this page compares specs only. Run the same three or four prompts through each and judge the results.

What is the difference between P-Image-OSS and MiniMax H3 Image?

P-Image-OSS is an image model from Pruna (Nov 3, 2026), taking text prompts and reference images and producing images. MiniMax H3 Image is an image model from MiniMax (Nov 10, 2026), taking text prompts and reference images and producing images.

Which is newer?

P-Image-OSS Nov 3, 2026; MiniMax H3 Image Nov 10, 2026.

Which can edit an existing image?

Takes a reference image as input: P-Image-OSS yes; MiniMax H3 Image yes.

Can I run any of these on my own hardware?

All open weights. Downloadable weights (MiniMax H3 Image Community License).