Resources · n8n integration

n8n MiniMax nodeConfigure MiniMax in n8n.

One credential, and a workflow can speak, draw and film. The n8n MiniMax node packs 5 operations over 4 resources: audio, image, text and video. It runs like any action node, once per incoming item, so the same record can leave with a voice-over, an illustration and a short clip.

Try the MiniMax node on n8nSee the 5 operations

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n MiniMax node actually do?

The MiniMax node connects a workflow to MiniMax AI models. It turns a text prompt into an image, a script into synthesized speech, a prompt or a first frame into a video, and a message into a model answer. Four resources, 5 operations, and the same credential behind all of them. Nothing else on the node: no trigger, no sub-node port, no list or search call.

Think of it as a media worker sitting in the middle of a workflow. Something upstream produces a row, a ticket or a form answer, MiniMax produces the asset, and a downstream node files it away. That last step is where the node pairs with the rest of your stack of connectors, most often Google Sheets for the tracking table or Slack for the review channel.

First scenario, a product catalog that needs visuals. A loop reads the product list, image.generate builds one picture per row from the description, and Number of Images set higher returns several variants in a single call so a human picks the best one. With Seed filled in, the same prompt gives the same picture again, which matters when someone asks for a regenerate with one word changed.

Second scenario, an audio version of written content. audio.textToSpeech takes the text field of the item, applies a voice and returns either binary audio or a URL, depending on Download Audio. Keep the binary output when the next node uploads a file, keep the URL when you only store a link. Sending the result out by email is a one-node hop away with Gmail.

Third scenario, short video from a still. video.imageToVideo animates a first frame, and on the models that support it a last frame or a face reference keeps the result consistent. Pair it with video.textToVideo when there is no image to start from.

When to reach for something else: the node talks to MiniMax models only. For an assistant that decides which tool to call, plans steps and keeps a conversation in memory, the AI Agent node is the root node and a chat model plugs into it as a sub-node, which is a node that never runs on its own. Vendors like OpenAI or Google Gemini cover that role too. The MiniMax node is the direct call: one item in, one asset out.

Known limits, worth checking before you build: the node is at version 1, the prompt fields are capped, and video duration is a choice between two values rather than a free number. Generation happens on the MiniMax side, so the workflow waits for the model, and a long video is not an instant response.

Connect

How do you connect MiniMax to n8n?

  1. 01

    Create the MiniMax credential

    Open the Credentials menu in n8n and add a MiniMax credential, or create it straight from the node's credential field. A credential is the stored secret n8n reuses instead of asking for a key in every workflow. Once saved, every MiniMax node in every workflow of that instance can select it from the dropdown, and rotating the key later is a single edit rather than a hunt through your workflows.

  2. 02

    Or switch on Gateway credits

    On n8n Cloud there is a second route. Select Use Gateway credits in the node's credential field and the node runs without a MiniMax account at all, billed through your n8n Cloud credits. It is the fastest way to test the node before anyone signs up for anything. On a self-hosted instance the option does not apply, so plan for your own key there.

  3. 03

    Run one operation before wiring the rest

    Pick the cheapest call to prove the setup: text.message with a short prompt and Simplify Output left on. If the answer comes back, the credential is valid for the other three resources too, since they all sit behind it. Then switch the operation to the one you actually need and map the real fields with expressions such as {{ $json.prompt }}, which reads a field from the incoming item.

Actions

What can each MiniMax operation do?

The MiniMax node exposes 5 operations across 4 resources. For each one: the node as you configure it in n8n, the required fields, and our field notes.

Resources × operations matrix
ResourceGenerate an ImageGenerate Video From ImageMessage a ModelText to SpeechGenerate Video From Text
Audio✓
Image✓
Text✓
Video✓✓

Audio

1 operation
01

Convert text to speech

audio.textToSpeech

What you see in n8n

Notes & use cases

Sends text to a MiniMax speech model and gets spoken audio back, as binary data on the item or as a URL.

Key parameters

  • Text: the script to read, capped at 10000 characters, usually {{ $json.text }}.
  • Voice ID: the voice reference, such as English_Graceful_Lady, taken from the MiniMax documentation.
  • Model: six choices from speech-02-hd to speech-2.8-turbo, HD for quality, turbo for speed and languages.
  • Download Audio: on, the file rides along as binary; off, only the audio URL comes back.
  • Options: Emotion, Speed, Volume, Pitch, Language Boost and Audio Format, where WAV works in non-streaming mode only.
Use cases
voice a weekly summary for a team that listens on the commute.

Image

1 operation
02

Generate an image

image.generate

What you see in n8n

Notes & use cases

Turns a written description into one or several pictures, returned as binary files or as URLs.

Key parameters

  • Prompt: the description of what to draw, up to 1500 characters, often assembled from item fields.
  • Model: image-01, the image generation model of the node.
  • Aspect Ratio: eight framings with their pixel sizes, from 1:1 at 1024x1024 to 21:9 at 1344x576.
  • Number of Images: 1 to 9 per request, useful when a human picks from a batch.
  • Download Image: binary file or URL only.
  • Options: Prompt Optimizer rewrites the prompt before sending it, Seed makes a result reproducible with the same parameters.
Use cases
illustrate a blog draft before publication, or generate a visual per product line straight from the catalog description.

Text

1 operation
03

Message a model

text.message

What you see in n8n

Notes & use cases

Sends one or more messages to a MiniMax language model and returns its answer, raw or trimmed.

Key parameters

  • Model: seven options from MiniMax-M2 to MiniMax-M2.7, with highspeed variants.
  • Messages: the conversation itself, each entry pairing a Prompt with a Role, user to ask, assistant to set a tone or persona.
  • Simplify Output: on, the node hands back a short response object instead of the full raw payload.
  • Options: System Message, Hide Thinking to strip the reasoning, Maximum Number of Tokens (1024 by default), Max Tool Calls Iterations (15, or 0 for no limit), Output Randomness (Temperature) and Output Randomness (Top P).
Use cases
classify an inbound message into a category before routing it.

Video

2 operations
04

Generate video from text prompt

video.textToVideo

What you see in n8n

Notes & use cases

Produces a short clip from a written scene, with camera moves written into the prompt itself.

Key parameters

  • Prompt: the scene, up to 2000 characters; bracket commands such as [Push in] or [Pan left] drive the camera.
  • Model: four choices, MiniMax-Hailuo-2.3 and MiniMax-Hailuo-02 for realism and longer output, T2V-01-Director when the camera commands matter, T2V-01 as the standard.
  • Duration (Seconds): 6 or 10.
  • Resolution: 720P, 768P or 1080P, and what is available depends on the model chosen.
  • Download Video: the clip as binary data, or only its URL.
  • Prompt Optimizer: rewrites the scene description before generation.
Use cases
build a teaser from a campaign line, or animate a scene described in a brief without opening an editing tool.
05

Generate video from image

video.imageToVideo

What you see in n8n

Notes & use cases

Starts from a still picture and animates it.

Key parameters

  • Image Input Type: url or binary, and that choice decides which of the next two fields is required.
  • Image URL: the public address of the first frame, JPG, JPEG, PNG or WebP under 20MB.
  • Input Data Field Name: the binary field carrying the image when it arrives from an upload node rather than a link.
  • Model: six options, from I2V-01 to MiniMax-Hailuo-2.3-Fast.
  • Prompt: optional here, up to 2000 characters, same bracket syntax for camera moves.
  • Options: Last Frame Image Input Type, supported by MiniMax-Hailuo-2.3 and MiniMax-Hailuo-02 only, and Subject Reference Input Type, supported by MiniMax-Hailuo-2.3 only, for facial consistency.
Use cases
animate a product photo for a social post.
Need help

Need help automating MiniMax with n8n?

A person reads every message.

FAQ

MiniMax and n8n, the questions that come next

01Is the MiniMax node included in n8n?
Yes. The node belongs to the n8n AI package and ships with n8n, on n8n Cloud as well as on a self-hosted instance running the Community Edition under the Sustainable Use license. There is nothing to install and n8n charges nothing extra for it. What you do pay for sits on the other side: MiniMax bills its own API calls under its own terms, and a workflow that renders a video every hour is a workflow that calls MiniMax every hour. On n8n Cloud there is a middle path, Gateway credits, where the node runs on your n8n Cloud credits instead of a MiniMax account.
02What do you need for the n8n MiniMax node to work?
A MiniMax credential, created once in the Credentials menu and reused across every workflow on the instance, or Gateway credits on n8n Cloud selected in the node's credential field. Beyond that, the required fields of the operation you picked: Text and Voice ID for speech, Prompt for image and text-to-video, and for image-to-video either Image URL or Input Data Field Name depending on the input type. A required field left empty stops the node before it calls anything. The official credential page was not available when this brief was built, so check the node panel for the exact field names.
03What are the limits of the MiniMax node in n8n?
The node is at version 1 and covers 5 operations, so anything outside generation is out of scope: no listing of past jobs, no voice cloning, no account management. Prompt lengths are capped, at 1500 characters for images and 2000 for video, and speech tops out at ten thousand characters of text. Video duration is a choice between 6 and 10 seconds, resolutions depend on the model, and the last frame and subject reference options only work on some Hailuo models. Generation runs on MiniMax servers, so the workflow waits for the result rather than getting it instantly.
04When should you use the MiniMax node instead of an AI Agent?
Use MiniMax when the job is one call with a known shape: render this image, read this text, animate this frame. The operation is picked in advance, the fields are mapped, and the output lands in the item. Use an AI Agent when the model has to decide what to do, call tools, loop and remember the conversation. In that setup the agent is the root node and a chat model plugs into one of its ports as a sub-node. The two also coexist: an agent decides a visual is needed, and a MiniMax node further down the workflow produces it.
05n8n or Make for MiniMax generation?
It depends on where your data should live and how you want to pay. n8n can be self-hosted with Docker or npm, so prompts, generated files and credentials stay on infrastructure you control, and the same workflow also runs on n8n Cloud without changes. Make is hosted only, with no self-hosting option, and billing is counted per operation, which is predictable but grows with volume. On the building side, both are visual. Media workflows tend to generate a lot of steps per item, which is the part worth simulating on your own volumes before choosing.
Hack'celeration Lab

Get our weekly integration tips.

No spam. Unsubscribe anytime.

Same platform

Other docs you might need on this stack.

Browse all integrations