Issues

• test: pin the GGUF per-type dequant transient table

  invoke-ai/InvokeAI#9720 · Filter by author lstein opened on Oct 9, 2026

• GGUF dequant transient table is exact, not an upper bound

  invoke-ai/InvokeAI#9719 · Filter by author lstein opened on Oct 9, 2026

• GGUF dequant transient: numpy-fallback (IQ) types over-reserve device memory ~4x

  invoke-ai/InvokeAI#9718 · Filter by author lstein opened on Oct 9, 2026

• [bug]: Generate settings components section to handle diffusers vs checkpoints

  bug
  Bug

  invoke-ai/InvokeAI#9717 · Filter by author joshistoast opened on Oct 9, 2026

• [enhancement]: HF Links on Model Manager Items

  enhancement
  Feature

  invoke-ai/InvokeAI#9716 · Filter by author joshistoast opened on Oct 9, 2026

• [enhancement]: Annotations and comments on media

  enhancement
  Feature

  invoke-ai/InvokeAI#9715 · Filter by author joshistoast opened on Oct 9, 2026

• [enhancement]: Board and Projects sharing

  enhancement
  Feature

  invoke-ai/InvokeAI#9714 · Filter by author joshistoast opened on Oct 9, 2026

• Track model manager follow-up notes from PR #9669

  invoke-ai/InvokeAI#9712 · Filter by author JPPhoto opened on Oct 8, 2026
  

• Model settings edit can leave pre-edit cache entries serving requests (commit-before-invalidate race)

  invoke-ai/InvokeAI#9711 · Filter by author lstein opened on Oct 8, 2026

• test(minimax-h3): audio-only Ref2VA test does not guard the generated audio rows

  invoke-ai/InvokeAI#9710 · Filter by author lstein opened on Oct 8, 2026
  

• [enhancement]: Web page for weekly image challenge

  7.0.0enhancement
  invoke-ai/InvokeAI#9707 · Filter by author lstein opened on Oct 8, 2026
  

• MiniMax H3 LoRAs targeting top-level projections (audio_patch_proj etc.) are rejected

  invoke-ai/InvokeAI#9706 · Filter by author lstein opened on Oct 8, 2026
  

• Follow up on graph collector validation findings from #9686

  invoke-ai/InvokeAI#9695 · Filter by author JPPhoto opened on Oct 7, 2026
  

• [bug]: Prevent stale in-flight model loads from reentering cache

  7.0.0bug
  invoke-ai/InvokeAI#9674 · Filter by author JPPhoto opened on Oct 4, 2026
  

• Multi-user mode: v7 has no controls to share boards or workflows

  7.0.0bugfrontend
  invoke-ai/InvokeAI#9673 · Filter by author lstein opened on Oct 4, 2026

• FLUX.2 starter bundle installs gated, non-commercial Klein 9B and FLUX.2 [dev]

  bugmodel manager
  invoke-ai/InvokeAI#9670 · Filter by author lstein opened on Oct 4, 2026
  

• Model cache: evicting a VRAM-resident model from RAM leaves its weights pinned in VRAM while a node still references it

  7.0.0
  invoke-ai/InvokeAI#9667 · Filter by author lstein opened on Oct 3, 2026
  1

• [bug]: Floating-window movement persists viewport clamp on untouched axis

  bug
  invoke-ai/InvokeAI#9650 · Filter by author JPPhoto opened on Oct 2, 2026
  

• [enhancement]: Support multiresolution inspection of very large images

  invoke-ai/InvokeAI#9615 · Filter by author kvnloo opened on Oct 1, 2026
  1

• Canvas Shape tool creates a raster layer instead of filling the selected inpaint mask or regional guidance

  bugfrontend
  invoke-ai/InvokeAI#9616 · Filter by author lstein opened on Sep 30, 2026

• Canvas: layer settings (regional prompts, control model) are hidden below the tool settings

  bugfrontend
  invoke-ai/InvokeAI#9617 · Filter by author lstein opened on Sep 30, 2026
  1

• Gallery deletion does not check whether media is still in use

  bugfrontend
  invoke-ai/InvokeAI#9618 · Filter by author lstein opened on Sep 30, 2026

• Possible model-support issues found while documenting model families

  bug
  invoke-ai/InvokeAI#9619 · Filter by author lstein opened on Sep 30, 2026
  2

• [bug]: Z-Image turbo regional guidance.

  bug
  invoke-ai/InvokeAI#9612 · Filter by author Jokerlorian opened on Sep 30, 2026

• FP8 Storage is switched on at install for MXFP8 checkpoints, where it costs quality

  invoke-ai/InvokeAI#9620 · Filter by author Pfannkuchensack opened on Sep 30, 2026