VSDC 11.3 AI Video Generator Brings fal.ai Models Into the Editor

What changed in VSDC 11.3

VSDC, the Windows video editor from Flash-Integro, has added generative AI in version 11.3. The release adds an AI video generator, AI image generation and text-based AI image editing, and these tools sit inside the editing interface itself.1 The update also expands Garmin and DJI telemetry overlays, adds handheld-camera, lens-correction and Audio Spectrum templates, and improves the Key Editor and general stability.13

The dates differ slightly across the coverage. The software catalog VideoHelp lists 11.3 as the newest build with an October 9, 2026 date.6 VSDC's own feature page is dated October 12.1 The wire release went out from Sofia, Bulgaria on October 13.4 Most likely the build shipped quietly and the publicity came a few days later, which is a common pattern for smaller desktop software vendors.

The AI story matters most here, but probably not the way the headline suggests. VSDC did not build its own text-to-video model. It added a front end for someone else's catalog of models.

How the AI generator works

VSDC's documentation is open about where the AI comes from. The generator runs through fal.ai, a platform that hosts many AI models, and generated clips or images go straight into the user's project.1 To set it up, users create a fal.ai account, generate an API key, enter it under Options → Internet Connection, and then open the generation panel from a new AI button in the top-right corner of the editor.1

The workflow is the same for images and video. Users switch to video mode with a camera icon, choose a model, write a prompt, adjust the settings and press Generate. Finished clips go to the Resources window and can then be placed on the timeline.1 From there they can be trimmed, color-graded, captioned and combined with real footage using VSDC's usual tools.13 The image side adds an editing mode where text instructions can change lighting, swap backgrounds or fix details. It works on new generations and on images already in a project, launched from the Resources window.5

The sources describe the integration differently. The wire release only mentions a "third-party platform" offering several generative models and does not name fal.ai.3 VSDC's product page and FAQ name it directly.15 Both accounts agree on the business model: generation is billed separately according to the chosen model.3 VSDC shows an estimated cost before each image or video request is sent.5 When the FAQ asks whether AI generation is free, it answers that generation draws on the user's own fal.ai account.1

In practice, this is a "bring your own key" design. VSDC does not have to run GPUs or pay for model licenses. Users pay for generation separately from the $20 one-time Pro license that VideoHelp lists.16 For a budget editor, this lets VSDC offer frontier generative video without raising its price. The cost is extra setup that people used to subscription apps may find awkward.

What VSDC users get access to

The choice of fal.ai matters because of how many models it carries. An independent profile describes fal as a generative media platform with more than 1,000 production models for image, video, audio and multimodal generation, including FLUX, Veo 3, Kling, Wan, Seedream and Nano Banana.22 fal's own video catalog currently lists text-to-video and image-to-video models including MiniMax H3 Max, Alibaba's Wan 3.0, Kling V3 in Turbo Pro and native 4K versions, and PixVerse C1, which can generate native audio.21

A September 2026 provider comparison describes fal as the broadest aggregator. It lists Veo 3.1, Kling v3, Seedance 2.5, Wan 3.0, LTX-2.3, MiniMax H3 and Grok Imagine 1.5 among its main video models, and FLUX.2, GPT Image 2.5, Nano Banana and Recraft V4 on the image side.24 The same analysis cites fal's own survey: 50% of responding organizations used fal for image generation and 44% for video.24

Prices vary a lot by model, so VSDC's cost preview is useful. On September 23, 2026, Kling v3 Pro cost $0.112 per second without audio and $0.168 with audio. Veo 3.1 Lite at 720p ran from $0.03 per second, while full Veo 3.1 reached $0.40 per second with sound.24 A ten-second establishing shot could therefore cost well under a dollar or several dollars, depending on the model chosen.

The model-agnostic approach also protects VSDC from churn among AI vendors. According to the same comparison, OpenAI said in March 2026 that its Videos API and all Sora 2 models would shut down on September 24, 2026.24 An editor tied to one vendor's model would have to rebuild that feature. An editor sitting on top of an aggregator mostly just loses one option from a menu.

Where voice and multimodal fit

On AI voice synthesis, the release says less than the category label suggests. VSDC's announcement covers image generation, video generation and image editing. Neither the product page nor the wire copy mentions text-to-speech or voice cloning.13 The changelog lists the AI additions as only those three tools.6

The underlying platform covers more. fal hosts text-to-speech and voice-cloning models, including ElevenLabs TTS and Chatterbox, along with 3D generation.25 fal's documentation says some video models produce synchronized audio natively while others return silent clips.21 So some audio may reach VSDC timelines as a side effect of the video model a user picks, even though VSDC has not built a dedicated voice tool. A separate voice or narration panel looks like an obvious next step. That is speculation, though, and nothing in the 11.3 materials promises it.

The multimodal angle is more concrete on the image side, where text instructions edit existing pictures.5 It is also visible in the wider model catalog, which includes image-to-video and reference-to-video endpoints alongside text-to-video.21 Whether VSDC exposes those image-conditioned modes, or only text-prompted generation, is not clearly spelled out in its walkthrough. The walkthrough describes the video workflow as model choice plus text prompt.1

Telemetry and templates: VSDC's existing niche

The AI features get the headline, but telemetry is where VSDC has built a real niche. Version 10.1 added GPS data import from GoPro, Garmin, Insta360, Suunto, TomTom and smartphone tracking apps.15 Later versions added speed, altitude and compass widgets, Insta360 gyroscope and acceleration data, and telemetry-sync markers.19

Version 11.3 adds three Garmin fields for Pro users: cycling cadence, heart rate and compass heading.1 DJI support now covers route, speed and altitude for drone and action-camera footage.13 VideoHelp's changelog lists coordinates rather than route for DJI, which is a small difference in wording.6 New telemetry templates fill themselves in once a data file is dropped on the timeline.5 The paid GPS + Telemetry package in the VSDC Store also gains more metrics.3

The creative templates aim at the same users. Handheld-camera presets add subtle camera shake to static or heavily stabilized shots. Wide-angle correction presets reduce fisheye distortion from action cameras. A new Audio Spectrum set offers 21 styles, each in five color variations.3

Workflow changes round out the release. The Key Editor gains multi-keyframe selection, dragging of whole curves, two vertical scrolling modes and one-click auto-scaling. Projects now remember open tabs, and VSDC can create a recovery project if Windows shuts down mid-session.1 The changelog also lists Shift-constrained movement, proportional resizing, Ribbon customization and better Lens Distortion accuracy.6

The takeaway

VSDC has added AI before. Its Pro tier already listed an "AI art generator,"18 and an earlier version added AI-model segmentation for cutting out objects.19 Version 11.3 is a different kind of move. Instead of offering one built-in model, VSDC now acts as a front end to one of the largest commercial model catalogs available.

The main reason this matters is how the work flows. With generation built into the editor, a missing opening shot or background becomes a prompt rather than a trip to a separate web app followed by downloading and importing the result. VSDC frames the feature exactly this way, as a way to create a missing visual without leaving the project.1 Users pay per generation and manage their own API key. In return, they can reach new models from a $20 editor as soon as fal adds them. For hobbyists and action-sports creators, who are VSDC's core audience, that is a practical way to add AI generation, and other low-cost editors are likely to follow the same approach.