--- title: P-H3-Image emoji: 🖼️ colorFrom: purple colorTo: indigo sdk: gradio sdk_version: 6.30.0 python_version: "3.12" app_file: app.py pinned: false startup_duration_timeout: 1h short_description: Text-to-image and image editing, based on MiniMax-H3 tags: - ml-intern models: - PrunaAI/P-H3-Image --- Built with ML Intern P-H3-Image on ZeroGPU (`xlarge`: a full RTX PRO 6000 Blackwell, 96 GB) — now behind two bespoke surfaces modeled on [briaai/Fibo-2.0-editing](https://huggingface.co/spaces/briaai/Fibo-2.0-editing), with the backend (`p_h3_image_fast.py`, prompt upsampler, warmup) unchanged: - **`/` — Studio** (`gr.Server` API + custom UI, Pruna design, dark and light themes): Generate and Edit modes with prompt upsampling, aspect-ratio and resolution controls (4 / 8 / 16 MP text-to-image, 4 / 8 MP edits at the source's aspect ratio), up to 8 reference images, seed control, all of the original examples, before/after split view, run history with one-click restore, and the upsampled prompt after each run. Both modes call the same `_render` as the classic UI did. - **`/workflow` — Workflow canvas** (`gr.Workflow`): node-based generate/edit pipelines you can rewire and chain — text-to-image into chained edits, or your own source image. To change the starter graph, edit `workflow.json` in the repo. The classic two-tab form UI is replaced by the studio; `sdk_version` moves to 6.30.0 for `gr.Server`/`gr.Workflow`.