Kandinsky 6.0 Pro
Kandinsky 6.0 Pro is the 29-billion-parameter model in Kandinsky 6.0, the video family Kandinsky Lab open-sourced under the MIT licence on 6 October 2026. It makes five-second clips with synchronized 44 kHz audio, lip-sync included, from text or a first frame, at a 480p base size that a separate super-resolution model lifts to 1080p or four times the size. fal serves it with that upscaler as an option and bills per clip by step count: $1.35 at the default 50 steps, $1.40 from an image. We ran it at the native 480p without upscaling, so these clips are smaller than the 720p most of the roster returns.
Anvisha Pai, Co-founder & CEO, Voyager
Kandinsky Lab · released
›What we asked for
A self-portrait of Kandinsky 6.0 Pro, Kandinsky Lab's video model: a painter's brush sweeps bold circles and lines of colour across a white canvas in a sunlit Moscow studio, the abstract shapes beginning to float off the canvas, the soft scrape of bristles and a quiet piano.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
At a glance
- Price
- $1.35 for a five-second 480p clip on fal at the default 50 steps, $1.40 from an image, read October 22, 2026. Upscaling to 1080p raises it to $1.53; 20 steps cost $0.56.
- Measured wait
- 14 to 26 minutes from queue submission to a five-second clip, across nine prompts in one batch, as slow as Seedance 2.5 at its slowest. Not a service guarantee.
- Length
- Five seconds; there is no duration setting.
- Open weights
- MIT licence, 29 billion parameters, on Hugging Face with a distilled variant; runs in Diffusers and ComfyUI.
- Audio
- Synchronized 44 kHz sound including lip-sync, on by default.
Other Kandinsky models: Kandinsky 6.0 Lite. Compare all 2.
Anvisha's take
Kandinsky 6.0 Pro is the open-weight model with the best manners on our product photo and the longest waits on the roster. Every clip took 14 to 26 minutes to come back from fal. In image-to-video it kept our handleless dripper exactly and did something no other model here thought of: a second, glass dripper full of grounds lowers into frame above it and coffee drips through into ours. The speaking woman is a natural handheld selfie that ends on a smile, the harbour aerial is steady and dense with boats, and the storm is a flat-colour print with a wave breaking on the rocks, though under a calm sky. It followed every prompt, which its Lite sibling did not. The misses: the title came back as 'MODA SUWWER CESSTIONES', the turntable product is a lidded jug that ends with two handles, and the spill starts with no ice and grows round cubes from nowhere. At $1.35 for five seconds of 480p it is one of the dearer clips here; its case is the MIT licence, not the hosted price. These are frame-based observations; audio was present on every clip but not scored by ear.
Kandinsky 6.0 Pro real output: the same 9 clips every model gets
Every video model on this site gets the same 9 clips, so you can compare like with like. These are the clips Kandinsky 6.0 Pro returned, one try each, nothing re-rolled or retouched. Each tile shows one frame of the clip until you press play. Generated on . The 9 clips took 830 to 1553 seconds each and cost $12.20 in total.
A clean product shot that holds its shape through a full slow rotation.
›What we asked for
A matte black ceramic pour-over coffee dripper rotating slowly on a white turntable, soft studio light, seamless white background, product video, no text.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
A face, lip sync and a spoken line, where the model makes audio.
›What we asked for
A woman in her 40s with short grey hair stands in a bright kitchen, looks at the camera and says 'Good morning, the coffee is ready', natural window light, handheld feel, she smiles at the end.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
Set text that stays spelled correctly while it moves.
›What we asked for
A title card reading 'MODA SUMMER SESSIONS' animates in over slow-motion footage of a festival crowd at dusk, then a smaller line 'Friday 12 June' fades in beneath it. Bold condensed type, tangerine on navy.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
A steady drone push-in with stable geometry.
›What we asked for
Drone shot pushing slowly forward over a coastal town at sunrise, terracotta roofs, fishing boats in the harbour, long shadows, cinematic and steady.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
A spill in slow motion: does water behave like water.
›What we asked for
A glass of water tips over on a wooden table and spills in slow motion, water and three ice cubes sliding across the wood, macro lens, morning light.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
Hands, a kettle and a steady pour without extra fingers or teleporting objects.
›What we asked for
Close-up of hands pouring hot water from a gooseneck kettle in slow circles into a pour-over coffee dripper, steam rising, shallow depth of field, warm cafe light.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
Many people and objects moving at once without merging.
›What we asked for
A crowded Saturday farmers market seen from slightly above, shoppers with tote bags, a dog on a leash, a busker with a guitar, bunting swaying, late morning light, gentle camera drift.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
A named style held across every frame rather than drifting to photoreal.
›What we asked for
A lighthouse on a rocky cliff in a storm, in the style of a Japanese ukiyo-e woodblock print, waves crashing in flat colour with visible wood grain, animated like moving paper cutouts.
fal-ai/kandinsky6-pro/text-to-video
aspect_ratio: 16:9, generate_audio: true
returned 864 by 480 pixels
Bring a still product photo to life without changing the product.
›What we asked for
The coffee dripper slowly fills as coffee drips through it, steam rising, the camera drifting gently around it, the dripper itself unchanged. (starting from The Nano Banana Pro product shot from the image fixture set.)
fal-ai/kandinsky6-pro/image-to-video
aspect_ratio: auto, generate_audio: true
returned 640 by 640 pixels
How the cost under each picture was worked out: fal's flat $1.35 per five-second 480p Kandinsky 6.0 Pro text-to-video clip at the default 50 steps, no upscaling (fal.ai/models/fal-ai/kandinsky6-pro, read 2026-10-22); fal does not return the charge.
Compare Kandinsky 6.0 Pro
Kandinsky 6.0 Pro pricing
| Where | Unit | Price | Source |
|---|---|---|---|
| fal | Five-second 480p clip, text-to-video At the default 50 steps: 20 steps is $0.56, 8 steps $0.24. fal's pricing API reports $1.35 per unit. | $1.35 | fal model page, read |
| fal | Five-second 480p clip, image-to-video | $1.40 | fal model page, read |
| fal | Five-second clip upscaled to 1080p Four times the size is $1.91. | $1.53 | fal model page, read |
Is Kandinsky 6.0 Pro free?
The weights are free under the MIT licence. On fal it is paid per clip.
- •GitHub and Hugging Face: code and weights for Pro and its distilled variant, plus the super-resolution model.
- •fal: per clip by step count, no free allowance found.
Kandinsky 6.0 on GitHub, read
Where to use Kandinsky 6.0 Pro
- fal, text-to-video · Hosted API and playground; the endpoint used for eight of the clips below.
- fal, image-to-video · First frame, centre-cropped to the nearest supported ratio; used for the image-to-video clip.
- Open weights · Kandinsky Lab's code and links to the Hugging Face weights, under MIT.
- Voyager · Runs inside Voyager, the agent for creative work, with your own fal key.
Voyager
Run Kandinsky 6.0 Pro inside Voyager
Voyager is an agent for creative work that runs every model in this reference, Kandinsky 6.0 Pro included. Brief it, and it picks the model, makes the thing and hands back something you can edit. Create an account to get started.
What we noticed running Kandinsky 6.0 Pro
- •Kept the photographed dripper exactly; a second, glass dripper of grounds lowers into frame above it and coffee drips through into ours. See the image to video clip.
- •The title animates in as 'MODA SUWWER CESSTIONES' over 'Frday 12 Uoy'. See the animated title card clip.
- •Seen from above, a lidded black jug turns on a drum-like stand and ends with two handles. See the product turntable clip.
- •The glass starts with no ice, tips and rolls, and round cubes appear from nowhere in the puddle. See the physics clip.
- •A handheld selfie that talks, glances away and ends on a smile: natural, if small at 480p. See the person speaking clip.
Kandinsky 6.0 Pro limits and API parameters
| Limit | Value |
|---|---|
| Duration | Five seconds at 480p; 1080p or four times the size through the optional upscaler. Kandinsky 6.0 on GitHub, read |
| Output size | Our 16:9 clips came back at 864 by 480, 24 frames a second, 5.06 seconds long; the square product photo gave a 640 by 640 clip. fal API schema, read |
›Every setting, for developers
fal (fal-ai/kandinsky6-pro/text-to-video)
fal API schema, read
| Parameter | Type | Values | Default |
|---|---|---|---|
| prompt | string | Free text Required; describes the video and its sound. | |
| negative_prompt | string | Free text Optional; defaults to the model's built-in negative prompt. | |
| aspect_ratio | enum | 16:9, 9:16, 1:1, 4:3, 3:4 | 16:9 |
| generate_audio | boolean | true, false | true |
| num_inference_steps | integer | 8 to 50 The price follows the step count. | 50 |
| guidance_scale | number | 1 to 10 | 5 |
| upscale_factor | enum | 2, 2.25, 4 Optional super-resolution; 2.25 turns 480p into 1080p. Not used for our clips. | |
| enable_prompt_expansion | boolean | true, false Rewrites the prompt with the model's own Qwen2.5-VL encoder. | false |
| seed | integer | Any Random when not set. |
Kandinsky 6.0 Pro image-to-video
fal API schema, read
| Parameter | Type | Values | Default |
|---|---|---|---|
| image_url | URL | Image Required; the first frame. | |
| aspect_ratio | enum | auto, 16:9, 9:16, 1:1, 4:3, 3:4 auto picks the supported ratio closest to the image and centre-crops it. | auto |
What Kandinsky 6.0 Pro will not make
fal runs a safety checker on the hosted endpoints that only authorized callers can relax; none of our prompts was refused. The open weights are MIT licensed, so your own deployment carries only the policies you add.
fal API schema, read
Alternatives to Kandinsky 6.0 Pro
Frequently asked questions
Is Kandinsky 6.0 Pro open source?
Yes. Kandinsky Lab released the code and weights for Pro (29B), Lite (3B), their distilled variants and a super-resolution model under the MIT licence on 6 October 2026.
How much does Kandinsky 6.0 Pro cost on fal?
$1.35 for a five-second 480p clip at the default 50 steps, $1.40 from an image, read October 22, 2026. Fewer steps cost less ($0.56 at 20), and upscaling to 1080p raises it to $1.53.
Why is Kandinsky 6.0 Pro so slow?
On fal our clips took 14 to 26 minutes each, queue included. The hosted endpoint runs the full 29B model at 50 steps by default; fal lets you lower the step count, which also lowers the price.
How this page is made
Prices, limits and settings are read from the linked vendor and host pages on the dates shown. Every picture was generated by us through fal with the request shown under it, and the original files are kept unedited. A generation that failed is shown as a failure. Model and vendor names are used to identify the products; no vendor imagery appears on this page.
