LoginSign Up
← Back to all sparks
I

InvokeAI

AI-ASSISTANTS
Velocity6.3

Open-source generative AI image creation and workflow studio

invoke.com ↗

InvokeAI 6.14 adds video generation via Wan 2.2 and native multi-GPU support.

generative-aivideo-generationlocal-inferencemulti-gpumodel-support
◆Current state
InvokeAI 6.14.0 shipped August 25 as the product's largest model-support expansion to date: Wan 2.2 video generation (text-to-video and image-to-video), Flux.2 Dev with 4K super-resolution via Flux.2 PiD, Krea.2-Turbo, Ernie Turbo, Ideogram 4, Anima controlnets, native Intel XPU support, and multi-GPU rendering. The 6.14.1 patch that followed two weeks later added workflow screenshots and gallery middle-click navigation.
◆Where it's heading
InvokeAI is expanding from an image-generation platform into a multi-modal local AI studio, adding video as a first-class output type alongside deepening hardware breadth (multi-GPU, Intel XPU, FP8). The pattern of rapid model additions — four major models in one release — indicates the team is prioritizing compatibility surface over depth, positioning InvokeAI as the widest-coverage self-hosted alternative to hosted generation APIs.
◆Prediction
Expect a 6.15 cycle to focus on video workflow integration (timeline editing, multi-clip sequencing) now that the generation primitive is in place.

◆Recent moves

  1. 8d ago

    InvokeAI 6.14.1: Workflow screenshots and gallery middle-click

    6.14.1 patches post-launch bugs and adds two quality-of-life features that missed the main release: workflow screenshots exportable as PNGs and middle-click gallery navigation. Incremental polish on top of the 6.14 expansion.

    View source ↗
  2. 20d ago

    InvokeAI 6.14.0

    ⚡ SPARK

    6.14.0 is the release that pushes InvokeAI from image tool to multi-modal local studio: Wan 2.2 video generation ships as a first-class feature alongside Flux.2 4K super-resolution and native multi-GPU support. This is the directional move the RC cycle was building toward.

    View source ↗
  3. 29d ago

    InvokeAI 6.14.0 RC2 (pre-release)

    RC2 of 6.14.0 is a pre-release preview of the features that shipped in GA — its content was fully superseded within ten days.

    View source ↗
  4. 1mo ago

    InvokeAI 6.14.0 RC1 (pre-release)

    RC1 was the first public preview of 6.14's video and multi-GPU features, but it's a pre-release iteration — the GA release is what matters for trajectory.

    View source ↗
  5. 2mo ago

    InvokeAI v6.13.6

    v6.13.6 fixes a crash in Qwen Image models introduced the previous day — a patch for a patch, invisible to users who didn't hit the crash.

    View source ↗
  6. 2mo ago

    InvokeAI v6.13.5

    v6.13.5 is a stability release with library updates and bug fixes, bridging the gap between 6.13.0's model additions and the 6.14 cycle's larger ambitions. No new user-visible capabilities.

    View source ↗