Bonsai-27B-GGUF
lmstudio-community · Kit
A GGUF conversion of the 27B Bonsai model for LM Studio, with a 3.80 GB Q1_0 language file and a 0.93 GB BF16 vision projector. Tested for text, vision and par…
An n8n sub-workflow that another workflow calls to make one video on Kie.ai’s Veo model. It posts a prompt, a frame and reference images, checks the task every five seconds, and returns the first video address as resultUrls.
Kie.ai Veo video sub-workflow is an n8n workflow in the Developer category, listed by Mindket. It is free. This is version 2, last updated on . Download it on this page or copy its JSON, then import it into an n8n canvas. Nothing runs until you activate it in n8n.
Called by another workflow
It starts when another workflow calls it through an Execute Workflow node, with four inputs:
The address of a finished video, returned to the calling workflow as resultUrls
What you need before you start.
Video generation
Makes the video from the prompt and images through its Veo generate endpoint, and reports the task’s status at record-info.
NeedsKie.ai API key
Checked by Mindket on 13 Nov 2026
This n8n workflow is a building block. Another workflow calls it to make one video on Kie.ai’s Veo model, and it hands back the video’s address. Seven nodes do the work. It receives a prompt, a frame shape, a list of image addresses and a generation type, and posts them to Kie.ai. Then it asks for the task’s status, waits five seconds and asks again until the video is ready, and returns the first address as resultUrls. Those are the four fields Mindket’s AI product photo to video generator sends to its own video sub-workflow, so this file can fill that slot.
n8n builders who make video workflows and want the Kie.ai call, with its waiting loop, in one place that several workflows share. Skip it if you want something to press: it has no form or button of its own and runs only when another workflow calls it.
| Industry | Who uses it | How they use it | Example |
|---|---|---|---|
| Online retail | Shop owners who automate their ads | Call it from the photo-to-video workflow, once per product | A shop’s workflow calls it for every new product photo. |
| Agencies | Automation freelancers | Keep the Kie.ai call in one place and reuse it across client workflows | A freelancer fixes an API change once, and three client workflows pick it up. |
| Social video | Teams that make Reels and Shorts | Send 9:16 as the frame and get a vertical clip | A team’s caller sets aspect_ratio to 9:16 for every run. |
| n8n development | Builders of long-running API calls | Copy the submit, check and wait pattern for another service | A builder reuses the loop for a different image API. |
It runs on the same n8n as the workflow that calls it, because an Execute Workflow node can only pick a workflow from its own n8n.
resultUrls, ready to download or send onGet it saves a zip with workflow.json and Mindket’s setup notes in your downloads folder. Its author puts setup at about 5 minutes.
| Need | Detail |
|---|---|
| n8n | On n8n Cloud or on a server of your own |
| Kie.ai | An account and its API key |
| A calling workflow | One with an Execute Workflow node, such as AI product photo to video generator |
workflow.json.Submit generation request and in Get request status, replace <YOUR_KIE_AI_API_KEY> in the Authorization header with your Kie.ai key. The brackets go too, so each header reads Bearer, a space and the key.The caller sends four fields. Each goes to Kie.ai under the name the request needs.
| Field | Type | What to send |
|---|---|---|
video_prompt | Text | The prompt for the video. It reaches Kie.ai as prompt |
aspect_ratio | Text | The frame, such as 16:9 or 9:16. It reaches Kie.ai as aspectRatio |
imageUrls | List of addresses | Addresses of the reference images, which Kie.ai must be able to open |
generationType | Text | Passed on as it is. The photo-to-video listing sends REFERENCE_2_VIDEO |
The file also fixes model at veo3_fast and sets enableFallback to true, both in the body of Submit generation request. Kie.ai’s own Veo pages explain what the fallback option changes.
A caller sends this for a product shot.
{
"video_prompt": "Cold brew bottle on a sunny café counter, slow dolly-in",
"aspect_ratio": "9:16",
"imageUrls": ["https://example.com/bottle.jpg"],
"generationType": "REFERENCE_2_VIDEO"
}The call comes back with one field, resultUrls, holding the first video address as text.
{ "resultUrls": "https://…" }If you are new. A sub-workflow is a workflow that other workflows call, the way a program calls a function. This one holds the slow part of making a video, which is sending the request and waiting for it. Test it from the calling workflow. Then open this one under Executions to see what each node received.
For experts. The status check runs once straight after the submit, with no wait, and later after each pause. Switch1 reads data.successFlag from Kie.ai’s reply.
| successFlag | Output of `Switch1` | What happens |
|---|---|---|
| 1 | Completed | Edit Fields keeps the first entry of data.response.resultUrls as resultUrls, and the call returns |
| 0 | Retry | Wait2 pauses, Edit Fields14 copies data.taskId forward, and Get request status runs again |
| 2 | Failed | No line leaves this output, so the run ends |
| 3 | Failed | The same |
Wait2 has no settings of its own, so it uses n8n’s default of five seconds for this version of the node. Change it there to poll less often.
Each generation costs money, and a failed task ends without a message.
An n8n sub-workflow. Another workflow calls it with a prompt, a frame, a list of image addresses and a generation type, and
it returns the address of a video that Kie.ai's Veo model made.
It takes the four fields that Mindket's "AI product photo to video generator" listing sends to its own video
sub-workflow, so it can be the one that listing asks you to supply.
workflow.json: the sub-workflow as n8n exports it, seven nodes. It holds no credential. Your Kie.ai key goes intoREADME.md: these notes.workflow.json.Submit generation request and Get request status. In each one's Authorization header, replace<YOUR_KIE_AI_API_KEY> with your Kie.ai key. The brackets go too, so the value reads Bearer, a space, then the key.| Field | Type | What to send |
|---|---|---|
| video_prompt | text | The prompt for the video. It reaches Kie.ai as prompt. |
| aspect_ratio | text | The frame, such as 16:9 or 9:16. It reaches Kie.ai as aspectRatio. |
| imageUrls | list of addresses | Addresses of the reference images, which Kie.ai must be able to open. |
| generationType | text | Passed on as it is. The photo-to-video listing sends REFERENCE_2_VIDEO. |
It returns one item with one field, resultUrls: the first video address, as text.
Submit generation request posts the four fields to https://api.kie.ai/api/v1/veo/generate, with model fixed atveo3_fast and enableFallback set to true.Get request status asks https://api.kie.ai/api/v1/veo/record-info for the task, straight away.Switch1 reads data.successFlag. A 1 goes to Edit Fields, which keeps the first entry ofdata.response.resultUrls, and the call returns. A 0 goes through Wait2 and Edit Fields14 back toGet request status. A 2 or a 3 has no line out, so the run ends.Wait2 has no settings of its own, so it waits n8n's default for this version of the node, five seconds. Change it there
to check less often.
enableFallback changes and what the other generation types need.README.md 3 KB workflow.json 9 KB
Updated by Mindket to the current version of this kit.
lmstudio-community · Kit
A GGUF conversion of the 27B Bonsai model for LM Studio, with a 3.80 GB Q1_0 language file and a 0.93 GB BF16 vision projector. Tested for text, vision and par…
LibertAIDAI · Kit
Runs DeepSeek-V4.1-Flash-1017 as a 2.9 bits-per-weight EXL3 quantization, a 196 GiB copy for people serving large models on their own hardware. Perplexity stay…
facebook · Kit
Generates short music clips from a text description or an audio prompt, in the small checkpoint of facebook's MusicGen. It suits developers and music hobbyists…
alpindale · Kit
Tags every character of C source code for syntax highlighting with a tiny neural network of about 984 weights, run through a small C program. For developers an…
prithivMLmods · Kit
Try a set of LoRAs for the FLUX.2 Klein image model in a Gradio app that runs in your browser or on your own computer. Made for people who make or test AI imag…
Sokheng · Kit
Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…