Issues

• [bug]: MiniMax-H3 video gen resolution locked to 768p only, need broader resolution options

  bug
  invoke-ai/InvokeAI#10241 · Filter by author NiceDragon8066 opened on Nov 11, 2026

• [Security] Require authentication no matter what.

  Feature

  invoke-ai/InvokeAI#10233 · Filter by author joshistoast opened on Nov 11, 2026
  

• [Feature] Add guidance to inpaint masks

  7.1.0
  invoke-ai/InvokeAI#10232 · Filter by author joshistoast opened on Nov 11, 2026

• [bug]: Model Manager incorrectly classifying Moss Image Qwen3.5-9B encoder as Text LLM model

  bug
  invoke-ai/InvokeAI#10217 · Filter by author NiceDragon8066 opened on Nov 9, 2026

• Hold the remaining T5/UMT5 encoder paths in float32 on CPU

  invoke-ai/InvokeAI#10215 · Filter by author Pfannkuchensack opened on Nov 9, 2026

• bug(graph): reversed If axes select the wrong For frame

  bug
  invoke-ai/InvokeAI#10207 · Filter by author JPPhoto opened on Nov 8, 2026
  

• [bug]: FLUX.2 Klein reference image fails with CUDA/CPU BFloat16 device mismatch

  bug
  invoke-ai/InvokeAI#10204 · Filter by author b092818627-rgb opened on Nov 8, 2026
  3

• Execution-engine fuzzer needs an immutable semantic baseline oracle

  bug
  invoke-ai/InvokeAI#10198 · Filter by author JPPhoto opened on Nov 7, 2026
  

• Nested For continuation fails in four reviewed graph shapes

  bug
  invoke-ai/InvokeAI#10197 · Filter by author JPPhoto opened on Nov 7, 2026
  

• fix(model-cache): adopting shared weights reserves full size and can evict the peer device's copy

  backendbugmodel manager
  invoke-ai/InvokeAI#10196 · Filter by author lstein opened on Nov 7, 2026

• [bug]: FLUX.2 DoRA layers on fused QKV projections are skipped at load

  bug
  invoke-ai/InvokeAI#10191 · Filter by author Pfannkuchensack opened on Nov 7, 2026
  2

• [bug]: SimpleTuner LyCORIS LoRAs (lycoris_ key prefix) are not identified on any base

  bug
  invoke-ai/InvokeAI#10174 · Filter by author Pfannkuchensack opened on Nov 5, 2026

• [bug]: OOM/error during generation leaves orphaned VRAM and subsequent 'Empty Cache' crashes with ASGI exception

  invoke-ai/InvokeAI#10170 · Filter by author skunkworxdark opened on Nov 5, 2026

• [enhancement]: Enhanced Workflow Sorting & Filtering (Library and 'This Project' views)

  invoke-ai/InvokeAI#10168 · Filter by author skunkworxdark opened on Nov 5, 2026
  

• Krea-2: read reference images with a GGUF encoder through the mmproj vision tower

  invoke-ai/InvokeAI#10154 · Filter by author Pfannkuchensack opened on Nov 5, 2026
  

• Run Diffusers and GGUF T5 encoders in float32 on CPU

  invoke-ai/InvokeAI#10144 · Filter by author Pfannkuchensack opened on Nov 4, 2026
  

• Per-device FP8 storage format breaks shared CPU weights on mixed-capability GPUs

  invoke-ai/InvokeAI#10142 · Filter by author Pfannkuchensack opened on Nov 4, 2026

• bug: Ideogram 4 frontend generation saves unused steps/cfgScale in metadata and does not recall sampler preset or overrides

  invoke-ai/InvokeAI#10140 · Filter by author skunkworxdark opened on Nov 3, 2026

• Deduplicate TAEHV decoder blocks between MiniMax H3 and the Tiny AutoEncoder package

  invoke-ai/InvokeAI#10118 · Filter by author lstein opened on Nov 2, 2026
  

• Krea-2: fp32 input to bf16 LayerNorm skips fused kernel on ROCm

  invoke-ai/InvokeAI#10117 · Filter by author lstein opened on Nov 2, 2026
  

• Wan I2V A14B without a reference image fails with a raw shape error

  invoke-ai/InvokeAI#10116 · Filter by author lstein opened on Nov 2, 2026
  

• [bug]: tied GGUF/SDNQ weights duplicate on device with RAM copies enabled

  bug
  invoke-ai/InvokeAI#10113 · Filter by author JPPhoto opened on Nov 2, 2026
  

• [bug]: Sidecar patching moves the cached inner layers of wrapper patches to the GPU for the whole patch scope

  bug
  invoke-ai/InvokeAI#10108 · Filter by author Pfannkuchensack opened on Nov 2, 2026
  

• [bug]: LoKr and LoHa layers with a DoRA magnitude apply without the magnitude (DoKr/DoHa)

  bug
  invoke-ai/InvokeAI#10107 · Filter by author Pfannkuchensack opened on Nov 2, 2026
  

• [bug]: patchmatch requires a successful GitHub API call at import time, so PatchMatch is silently disabled when the machine is offline

  bug
  invoke-ai/InvokeAI#10100 · Filter by author truekry opened on Nov 1, 2026