v3.3.1
Breaking Changes📦 localaiView on GitHub →
⚠ 1 breaking✨ 3 features🐛 1 fixes🔧 6 symbols
Summary
This minor release introduces support for Flux Kontext and Flux krea for image editing and generation. It also resolves critical bugs related to Intel GPU images and renames the corresponding container image tags.
⚠️ Breaking Changes
- Intel GPU container images have been renamed. \`latest-gpu-intel-f32\` and \`latest-gpu-intel-f16\` are replaced by a single image: \`latest-gpu-intel\`. For AIO images, \`latest-aio-gpu-intel-f32\` and \`latest-aio-gpu-intel-f16\` are replaced by \`latest-aio-gpu-intel\`.
Migration Steps
- If you were using the old Intel GPU image tags, update your Docker run commands to use the new unified tag: \`localai/localai:latest-gpu-intel\` (or \`localai/localai:latest-aio-gpu-intel\` for AIO).
- To use Flux Kontext image editing, install the model using \`local-ai run flux.1-kontext-dev\`.
✨ New Features
- Added support for Flux Kontext, enabling image editing via the API using models like \`flux.1-kontext-dev\`.
- Added support for Flux krea.
- Added support for loading LoRAs in stablediffusion-ggml.
🐛 Bug Fixes
- Fixed an important bug regarding Intel-GPU Images, including setting the GPU vendor on Intel images and cleanup (related to PR #5945).
🔧 Affected Symbols
localai/localai:latest-gpu-intel-f32localai/localai:latest-gpu-intel-f16localai/localai:latest-aio-gpu-intel-f32localai/localai:latest-aio-gpu-intel-f16flux.1-kontext-devstablediffusion-ggml