Skip to content

SDAI 1.1.0 - Android SDXL, iOS CoreML, Bonsai update

Latest
Latest

Choose a tag to compare

View all tags
@ShiftHackZ ShiftHackZ released this 19 Jun 08:44
· 1 commit to master since this release
1.1.0
8eeb37b
feature-graphic

SDAI 1.1.0

SDAI 1.1.0 is a feature release focused on local generation, provider choice, device insight, and practical usage tools across Android and iPhone.

Highlights

  • Added an on-device benchmark for local generation guidance.
  • Added Android Local SDXL generation through stable-diffusion.cpp.
  • Added iPhone local generation through Silicon Diffusion Core ML.
  • Added iPhone local generation through Silicon Diffusion PrismML Bonsai.
  • Added Fal.ai and ArliAI provider support.
  • Added storage usage and network usage screens.
  • Improved provider selection with search, filters, sorting, readiness labels, and feature tags.
  • Improved iOS gallery loading and background-work handling.

Local Generation

This release expands SDAI's local generation story on both platforms:

  • Android Local SDXL uses stable-diffusion.cpp for on-device SDXL-compatible text-to-image generation.
  • Silicon Diffusion Core ML brings Core ML Stable Diffusion generation to supported iPhones.
  • Silicon Diffusion PrismML Bonsai adds a beta PrismML Bonsai Image 4B / MLX local generation path for supported iPhones.

Local generation remains device- and model-dependent. The new benchmark is designed to help users understand whether their device is a good fit before committing to heavier runs.

Providers

  • Added Fal.ai hosted generation support for compatible text-to-image and image-to-image endpoints.
  • Added ArliAI SDNext-compatible hosted generation support, including checkpoint discovery, sampler selection, steps, CFG scale, negative prompt, and batch generation where supported.
  • Improved the provider setup experience with search, type/readiness/tag filters, sorting, and clearer provider labels.

Usage Tools

  • Added storage usage screens for cache, gallery, and downloaded model files.
  • Added network usage screens for model downloads, configuration downloads, and provider inference traffic.
  • Added local benchmark result storage and recommendation flow for supported local providers.

iOS

  • Added Core ML and Bonsai local-provider runtime integration.
  • Improved gallery loading performance.
  • Added iOS background-work support for long-running generation flows.

Android

  • Added stable-diffusion.cpp based Local SDXL provider integration.
  • Added model-catalog and local-model selection support for SDXL workflows.
  • Added benchmark and usage-tool platform actions.

Notes

  • Local AI generation can be memory-heavy. If a model is too large for the device, try a smaller or quantized model and check the benchmark recommendation first.
  • iOS local providers are intended for supported iPhone devices. This release does not announce iPad support.
  • SDAI remains open source, ad-free, telemetry-free, and provider-first.

Download and project links:

Assets 4
Loading