We’re always working to enhance your experience! Keep up with CSuite’s latest updates and new features. The app updates itself automatically, or you can get the latest version from the download page.
Alibaba's Wan 3.0 Prime joins the video catalog on Runware as the quicker tier of Wan 3.0, rendering a clip in roughly half the time for about 1.4 times the price. Everything else matches the base model: 2 to 30 seconds at up to 1080p, first and last frame control, reference media, and native audio.
ModelsWan 3.0 PrimeWan 3.0
Granite 4.2 joins the local models through Ollama in 3B, 8B, and 30B sizes, all three of them able to call tools and read a 128K context. The 3B is a roughly 2.2 GB download that runs on 8 GB of memory, so there is a size here for a modest laptop as well as for a workstation.
Every model list now opens with the most recent releases instead of an arbitrary order, in the model pickers, the provider lists under Models, and the workflow model node. When two models suit what you asked for equally well, Auto now picks the newer one.
Ornith 1.5 joins the local models through Ollama in a 9B and a 35B size. Both read images and call tools, so they work in chat and in workflows as well as in the text workspace.
When the model is set to Auto, you can type settings into the prompt and they are supplied for you: "red car, 16:9" picks the aspect ratio, and "8 seconds, with audio" sets a clip's length and sound. Only settings the chosen model actually supports are applied, and they count for that run alone rather than changing your defaults.
The wand that rewrites a rough prompt now takes the selected model and whatever you have attached into account. It keeps the subject of a song instead of flattening it into a mood, writes real lyrics on the models that sing from the prompt, turns into an editing instruction when there is an image attached, and describes motion rather than scenery when you have set a start frame.
Every model picker now has an Auto option that reads what you are asking for and chooses a model to match, so a logo goes to a model that returns a vector file and a quick draft goes to a fast one. New installs start on Auto everywhere, and when the best model belongs to a provider you have not connected, the app offers the best one you can actually run.
Alibaba's Wan 3.0 is now in the catalog through Runware, with clips from 2 to 30 seconds at up to 1080p, first and last frame control, reference images, and native audio.
ModelWan 3.0
Onboarding and the License panel now show your real license status and the reason behind it, an expired license can enter a new key, and Generate buttons explain in a tooltip why they are disabled instead of failing when you click them.
A full audit of the app fixed around ninety issues, from editing and file-handling edge cases to cloud generations being billed twice on a retry.
Four Recraft image models arrive, two of which are vector models that return real SVG files. Z.ai's GLM-5.2 joins the text models, and Veo 3.1 Lite adds a cheaper Veo tier on both providers.
ModelsRecraft v4.1Recraft v4.1 ProRecraft v4 VectorRecraft v4 Vector ProGLM-5.2Veo 3.1 Lite
Qwen Image 3 and 3 Pro and Grok Imagine Image 2.0 are now available on Replicate, and Kimi K2.6 and Grok Text-to-Speech on Runware, so the key you already have is more likely to reach the model you want.
ModelsQwen Image 3Qwen Image 3 ProGrok Imagine Image 2.0Kimi K2.6Grok Text-to-Speech
Asking a Gemini chat model on Runware for an image generated the image and then failed on the next round, leaving only an error in the thread. The round is fixed, and a failure later in a turn keeps whatever the turn already produced.
A .deb install cannot replace itself, so instead of trying, it now tells you when a new version is out and links straight to the download page.
The video AI menu can strip a background and write a transparent WebM, so a clip can sit over other layers in a composition. AI actions now sit at the top of that menu.
CSuite now ships a .deb alongside the AppImage, so it installs with a proper desktop entry and its own icon in the dock instead of a generic placeholder.
Once an update is downloaded, macOS shows a Restart to update pill in the title bar, and the app also checks for updates when you come back to the window.
Ollama removed image generation in a recent release, which broke local image models. The app now bundles a second runtime pinned to the last version that supports them, downloaded only when you install an image model.
The Stop button cancelled cloud work but left local runtime generations running until they finished. It now cancels both.
Describe the workflow you want and the assistant builds or edits the graph for you, the same way the composition chat edits a timeline. Each edit is a single undo step.
MiniMax H3 on Replicate, Gemini 3.6 Flash on Runware, and Qwen3.8 27B for local use through Ollama.
Reference images, videos, and audio are now wired across every model and both cloud providers, with a picker in each composer that shows exactly what the selected model accepts.
PDF, text, Markdown, and HTML files can be attached to a chat and their text is read into the conversation. Cloud models that support vision now accept images too.
Chat can run one of your saved workflows by name and post the results back into the thread as attachments.
Title clips can use any font installed on your machine, searchable in the picker, and the composition assistant knows which fonts are available when it writes a title.
Stop now cancels the work in progress instead of only hiding it, every round has a timeout, and you can copy a single message or the whole conversation as Markdown. Replies can be regenerated, and your last message can be edited and resent.
Runs cancel for real, a failed node keeps its error and offers a Retry, the toolbar estimates cost before you run, and there is auto-layout, pre-run validation, nine new edit operations, and the option to save your own graph as a template.
The dashboard now opens with six create tiles and a Files panel grouped by type, including compositions, workflows, and chat threads. Clicking any card opens the item in the workspace it belongs to.
A still on the composition timeline can be sent to a video model and replaced by the result without leaving the editor.
Lightricks LTX-2.5 joins the video catalog in both a fast and a pro tier.
ModelsLTX-2.5 FastLTX-2.5 Pro
The rebuilt editor reached parity and the previous video editor was removed. Compositions made in the old editor are not carried over, so anything you still need from one should be exported first.
Switch a composition between square, landscape, and portrait and every frame reflows to the new shape.
Grok Imagine Image 2.0 joins the image models, with Muse Glimmer and Nemotron 3.5 Lightning available locally through Ollama. Runware image models moved to their 2K tier.
ModelsGrok Imagine Image 2.0Muse GlimmerNemotron 3.5 Lightning
Compositions now have first-class tracks you can lock, mute, and hide, drag and drop onto the canvas, direct manipulation of clips, an AI chat that edits the timeline through real undoable operations, asset generation from that chat, and MP4 export with a choice of resolution and framerate.
ByteDance's Seedance 2.5 is available on both Replicate and Runware.
ModelSeedance 2.5
Trimming an audio file appeared to have no effect because the result was written in a different format than the one on screen.
If the app or that workspace is not in front, a finished generation now raises a system notification, and clicking it takes you straight there. There is an on and off switch in Account.
Pressing Stop on a Runware generation left the button spinning instead of ending the run.
Background removal returned a 404 because the underlying model had moved. Community tool models are now pinned to a known version.
Several Runware voice models failed because no voice was being sent, and the music models failed the same way. Both now carry the defaults the provider expects.
The wand that rewrites a rough prompt into a fuller one is now in the text and video workspaces as well as image and audio.
Flux 3 Video, MiniMax H3, Qwen Image 3 and Qwen Image 3 Pro, and DeepSeek V4 Pro.
The whole app now has a light theme. It follows your system setting by default, and Account has a control if you want to pin it to light or dark.
Video clips can be trimmed frame accurately, played back between half and double speed with the pitch preserved, stripped of audio or have their audio extracted to the Audio workspace, and exported as a GIF.
The transport bar gained a loop toggle, preview speed, and single frame stepping with the arrow keys. Any frame can be saved as an image or used as the start frame for a new generation.
An info popover shows duration, resolution, frame rate, codecs, and how the clip was generated, and the AI menu can upscale a clip to 1080p or 4K. A clip can also be appended straight to a composition.
The image toolbar can copy the picture straight to the clipboard and flip it horizontally or vertically, and resizing offers ready-made sizes for Open Graph, Instagram, Stories, and X that crop to fit rather than stretch.
Each generated image remembers how it was made. The info popover shows the model and prompt and can refill the composer with both, and your recent prompts sit as chips above it.
Images can be adjusted for brightness, contrast, and saturation with a live preview, and the annotate tool now has brush, arrow, rectangle, text, and eraser with per-step undo.
An AI menu on the image toolbar upscales two or four times and removes backgrounds, generation shows staged progress instead of a single spinner, and the composer shows roughly what an image will cost before you run it.
No subscriptions. No seats. No renewals. Buy CSuite once, future updates included.
Secure checkout via Stripe. Already have a license? Download the app