Feat[model support]: Qwen Image — full pipeline with edit, generate LoRA, GGUF, quantization, and UI#9000

Merged

lstein merged 60 commits intoinvoke-ai:mainfrom

lstein:feat/qwen-image-edit-2511

Apr 12, 2026

Commits on Mar 28, 2026

refactor: rename QwenImageEdit -> QwenImage base type

lstein
and
claude
committed
feat: add QwenImage variant type to distinguish edit vs txt2img models

lstein
and
claude
committed
fix: make QwenImage variant optional to fix model detection tags

lstein
and
claude
committed
fix: rename edit starter model variables to qwen_image_edit_* prefix

lstein
and
claude
committed
fix: remove unused exports (zQwenImageVariantType, isQwenImageEditMainModelConfig)

lstein
and
claude
committed
fix: restore correct GGUF filenames in Qwen Image Edit starter model URLs

lstein
and
claude
committed
feat: tag edit starter models with variant, pass variant through install

lstein
and
claude
committed
fix: only set zero_cond_t=True for edit-variant GGUF models

lstein
and
claude
committed
fix: prevent Flux LoRAs from being detected as Qwen Image LoRAs

lstein
and
claude
committed
test: add Qwen Image LoRA model identification tests

lstein
and
claude
committed
chore: ruff
lstein
committed
chore: ruff
lstein
committed
fix: don't force reference image to output aspect ratio in VAE encoding

lstein
and
claude
committed

Commits on Mar 29, 2026

Commits on Apr 1, 2026

Commits on Apr 5, 2026

Commits on Apr 7, 2026

Commits on Apr 10, 2026

Commits on Apr 12, 2026