Resources · Make integration

OpenAI (ChatGPT, Whisper, DALL-E) Make integrationAutomate OpenAI (ChatGPT, Whisper, DALL-E) with Make.

Can Make run ChatGPT, Whisper and DALL-E for you? Yes. The OpenAI (ChatGPT, Whisper, DALL-E) Make integration ships 42 modules for text, images, audio, video and batch jobs. Here you connect your account and build a first scenario, step by step.

Verified Trustpilot reviews · AI, automation & growth agency

What you can automate

What does the OpenAI (ChatGPT, Whisper, DALL-E) Make integration do?

It lets Make send work to OpenAI models and bring the result back into your other tools, without writing code. In Make, that automation is called a scenario: a chain of modules, each module being one brick (a trigger, an action or a search). OpenAI gives you 42 of them: 2 triggers, 35 actions and 5 searches.

Support replies drafted before anyone opens the ticket. A new ticket arrives in your helpdesk, Generate a response writes a first answer, and your agent only edits and sends. Add Generate a moderation in front if the text comes from the public.

Call recordings turned into notes. A sales call lands as an audio file, Generate a transcription turns it into text, then Transform text to structured data pulls out the company name, the need and the next step so they can go straight into your CRM.

Product visuals on demand. A marketer fills a row with a product and a mood, Generate images produces the picture with GPT Image, and Edit images reworks an existing shot instead of starting over.

What the integration does not do: it has no instant trigger. Both triggers are scheduled, which means Make checks OpenAI when the scenario runs rather than being pinged. For anything the 42 modules miss, Make an API call reaches the rest of the OpenAI API with the same connection. The same app also exists as a node in n8n, described on the OpenAI (ChatGPT, Whisper, DALL-E) n8n integration page; the n8n vs Make comparison helps you pick a tool. If you are new to the editor, the Make training covers the basics.

Connect

How do you connect OpenAI to Make?

  1. 01

    Add an OpenAI module

    In your scenario, click the +, search for OpenAI and pick any module. Make then asks for a connection: your OpenAI account linked to Make once and reused by every OpenAI module afterwards. Click Create a connection.

  2. 02

    Name it and paste your key

    Give the connection a name you will recognize later, for example the project or client it belongs to. Then authorize Make, either on the app's own page or by pasting the key OpenAI gives you, depending on what the window asks for.

  3. 03

    Save and check the fields load

    Click Save. Back in the module, the fields should now fill with what Make can read from your account. If they stay empty, the connection did not go through and it is worth recreating it before building further.

First scenario

Your first scenario: a text and its picture, ready on schedule

GoalWhen the scheduled time comes, Make asks OpenAI to write a text with Generate a response, then Generate images draws a matching picture.

  1. 01

    Create the scenario

    Open the Scenarios page, create a new scenario and click the big + in the middle of the canvas. That empty circle is where your first module goes.

  2. 02

    Add Generate a response

    Search for OpenAI, choose Generate a response and pick your connection. Type the instruction you would give ChatGPT, say a product teaser for your newsletter. This module writes the text.

  3. 03

    Add Generate images

    Click the + on the right and add Generate images. For the prompt, click the text from the previous module instead of retyping it: the picture then follows what OpenAI just wrote.

  4. 04

    Test with Run once

    Click Run once. Each module shows a bubble with the bundles it received, a bundle being one item passed along. Open them: you should read the text, then find the image.

  5. 05

    Schedule and switch on

    The clock on the first module sets when it runs: every 15 minutes by default, so slow it down if you need fewer drafts. Prefer another app to start it? Put that app's trigger in front. Then switch the scenario on.

Triggers

What can start an OpenAI scenario?

2 modules start a scenario when something happens in OpenAI (ChatGPT, Whisper, DALL-E). A scheduled trigger is checked by Make on the scenario's schedule; an instant one is called by OpenAI (ChatGPT, Whisper, DALL-E) as soon as it happens.

OpenAI (ChatGPT, Whisper, DALL-E)1Watch batch completed

Watch batch completed

Scheduled trigger

In MakeTriggers when a batch is completed.

Starts your scenario once a batch has finished on the OpenAI side. A batch is a pile of requests you hand to OpenAI in one go (with Create a batch) instead of calling the model one request at a time, and this trigger tells you the pile is done.

How it fires
scheduled. Make checks the app on the scenario's schedule, so a completed batch is picked up at the next run, not the second it ends.
When to use it
you sent thousands of product descriptions for rewriting and want the results pushed into your store as soon as the next check finds the batch closed.
Watch out
follow it with Get a batch to read the details rather than guessing them from the trigger alone.
OpenAI (ChatGPT, Whisper, DALL-E)2Watch video jobs

Watch video jobs

Scheduled trigger

In MakeTriggers when a video job is created.

Fires when a new video job appears in your OpenAI account. Video generation with Sora does not return a finished file straight away; it creates a job, and this trigger lets a separate scenario notice every new job without anyone watching the dashboard.

How it fires
scheduled. Make checks the app on the scenario's schedule and collects the jobs created since the last check.
When to use it
several people on the content team launch videos, and you want one log in a sheet or a message to the channel for each new job.
Watch out
it reacts to a job being created, not finished, so check its status with Get a video before sharing a link.
Modules

What can Make do with OpenAI?

OpenAI (ChatGPT, Whisper, DALL-E) gives you 40 modules. For each one: what it does for you, when to reach for it, and what to watch out for.

OpenAI (ChatGPT, Whisper, DALL-E)1

Add files to a vector store

Action

In MakeAdds files to a specified vector store or, if not specified, creates a new vector store based on the configuration.

Places documents into a vector store, the searchable library an OpenAI assistant consults before answering. Leave the store unspecified and OpenAI builds a fresh one from your settings instead.

When to use it
a new returns policy PDF is approved and your support assistant should quote it the next time a customer asks.
Watch out
an empty store field means a brand new store each run, so pick the existing one if you want a single library.
OpenAI (ChatGPT, Whisper, DALL-E)2

Analyze images (Vision)

Action

In MakeAnalyzes images according to specified instructions.

Looks at one or more pictures and answers the instruction you give it: describe, compare, read a label, spot a defect. The output is text you can route anywhere in the scenario.

When to use it
customers send photos of damaged parcels and you want a short written description added to the claim before a human reviews it.
Watch out
the quality of the answer follows the precision of your instruction, so say exactly what to look for.
OpenAI (ChatGPT, Whisper, DALL-E)3

Cancel a batch

Action

In MakeCancels an "in-progress" batch. The batch will be in status "cancelling" for up to 10 minutes, before changing to "cancelled", where it will have partial results (if any) available in the output file.

Stops a batch that is still running. The batch sits in a cancelling state for up to 10 minutes before it becomes cancelled, and any partial results are kept in its output file.

When to use it
someone spots a typo in the prompt after launching a large rewriting batch and you want to stop paying for wrong answers.
Watch out
the status does not flip at once, so a Get a batch placed right after may still read cancelling.
OpenAI (ChatGPT, Whisper, DALL-E)4

Create a batch

Action

In MakeCreates and executes a batch of API calls.

Hands OpenAI a whole set of API requests to execute together, rather than one module run per request. It is the way to process a large volume of prompts without a long chain of single calls.

When to use it
your catalog export has hundreds of rows that each need a new product title, and nobody needs the answers this very instant.
Watch out
the answers come back later, so pair it with Watch batch completed in a second scenario to collect them.
OpenAI (ChatGPT, Whisper, DALL-E)5

Create a skill

Action

In MakeCreates a new skill.

Registers a new skill in your OpenAI account, a reusable package of know-how your models can call on. This module creates the container; the content itself lives in its versions.

When to use it
your operations team maintains a set of house rules in a repository and wants each new rulebook to appear as its own skill without manual setup.
Watch out
plan the naming before automating it, because a scenario that runs twice will happily create two skills.
OpenAI (ChatGPT, Whisper, DALL-E)6

Create a skill version

Action

In MakeCreates a new immutable skill version.

Adds a new version to an existing skill. Versions are immutable: once created, one cannot be edited, you publish a newer one instead, which keeps a clean history of what changed.

When to use it
a manager approves an updated pricing guideline and you want that approval to produce the next version of the matching skill.
Watch out
creating a version does not make it the default; Update the default version pointer for a skill does that.
OpenAI (ChatGPT, Whisper, DALL-E)7

Delete a conversation

Action

In MakeDeletes an existing conversation.

Removes an existing conversation from OpenAI, with the exchanges stored in it. Use it to tidy up threads that your automations opened and no longer need.

When to use it
a customer closes their account and your offboarding scenario should erase the chat history created for them.
Watch out
deletion is final, so add a filter between modules to make sure only the right conversation reaches this step.
OpenAI (ChatGPT, Whisper, DALL-E)8

Delete a response

Action

In MakeDeletes an existing model response.

Erases a model response that OpenAI kept on its side. Only that stored answer disappears; what you already copied into a sheet or a CRM stays where it is.

When to use it
a draft generated for an internal review contained personal data and you want it gone from OpenAI once the review is logged.
Watch out
any later Get a response on that same answer will fail, so delete at the very end of the flow.
OpenAI (ChatGPT, Whisper, DALL-E)9

Delete a skill

Action

In MakeDeletes a skill. This action cannot be undone.

Removes a whole skill from your account. The action cannot be undone, which makes it the most sensitive module in the skills family.

When to use it
a product line is discontinued and the skill that described it should vanish along with the catalog entry.
Watch out
run List skills first and test on a throwaway skill with Run once, because a wrong mapping here costs you work you cannot recover.
OpenAI (ChatGPT, Whisper, DALL-E)10

Delete a skill version

Action

In MakeDeletes a skill version. This action cannot be undone.

Drops a single version of a skill while keeping the skill and its other versions. Like every delete in the skills family, it cannot be reversed.

When to use it
a draft version was published by mistake from a test scenario and you want the history cleaned without touching the version in use.
Watch out
check which version the default pointer targets before removing anything, or you may pull the rug from under your assistant.
OpenAI (ChatGPT, Whisper, DALL-E)11

Delete a video

Action

In MakeDeletes an existing video.

Removes a generated video from your OpenAI account. It pairs naturally with List video jobs when you want to clear out old or rejected renders.

When to use it
a reviewer marks a clip as rejected in your approval board, and that rejection should also clean the file on the OpenAI side.
Watch out
download or store any video you want to keep before this step, since the scenario cannot bring it back.
OpenAI (ChatGPT, Whisper, DALL-E)12

Download a container file

Action

In MakeDownloads a file from a container.

Fetches a file stored in an OpenAI container, the workspace where the model can create files while it works, and hands it to Make as a real file you can move.

When to use it
the model produced a chart or a spreadsheet during an analysis, and you want that file saved to your drive or attached to an email.
Watch out
the output is a file, so the next module has to be one that accepts files, like an upload or an attachment.
OpenAI (ChatGPT, Whisper, DALL-E)13

Download a skill

Action

In MakeDownloads a skill as a ZIP file using its ID.

Retrieves a whole skill as a ZIP file, located by its ID. You get the package itself, ready to archive or pass to another system.

When to use it
your compliance lead wants a copy of every skill stored outside OpenAI whenever one is changed.
Watch out
you need the skill ID; List skills or Get a skill earlier in the scenario gives it to you so nobody has to copy it by hand.
OpenAI (ChatGPT, Whisper, DALL-E)14

Download a skill version

Action

In MakeDownloads a skill version as a ZIP file.

Pulls one specific version of a skill as a ZIP, rather than the skill as a whole. Useful when you care about an exact snapshot.

When to use it
a customer asks which rules the assistant followed when it answered them, and you want to attach the version that was live at the time.
Watch out
pick the version on purpose; List skill versions shows the candidates so you do not archive the wrong snapshot.
OpenAI (ChatGPT, Whisper, DALL-E)15

Edit images

Action

In MakeCreates an edited or extended image given one or more source images and a prompt.

Takes one or more existing pictures plus a prompt and returns an edited or extended version: a new background, a wider frame, an added object.

When to use it
the product photos from your supplier are on white and your ad team needs the same items staged in a kitchen or on a desk.
Watch out
prefer Generate images when you have no source picture at all; this module is for reworking, not for starting from nothing.
OpenAI (ChatGPT, Whisper, DALL-E)16

Generate a completion

Action

In MakeGenerates a chat completion.

Sends a conversation to a chat model and returns its reply. It is the classic ChatGPT call: messages in, one reply out.

When to use it
an HR assistant should rewrite a job ad in a friendlier tone from a short list of messages you map from a form.
Watch out
for new builds, look at Generate a response too, which is the other text generator in the list; keep one style per scenario so mappings stay readable.
OpenAI (ChatGPT, Whisper, DALL-E)17

Generate a moderation

Action

In MakeQualifies whether the provided image or text(s) contains violent, hateful, illicit or adult content.

Checks a text or an image and tells you whether it contains violent, hateful, illicit or adult content. The result is a verdict you can filter on, not a rewrite.

When to use it
reviews or comments come from the public, and you want anything flagged kept away from your website and routed to a moderator.
Watch out
put a filter right after it; the module flags content but does not block anything on its own.
OpenAI (ChatGPT, Whisper, DALL-E)18

Generate a response

Action

In MakeGenerates a response.

Produces an answer from an OpenAI model and stores it as a response you can fetch or delete later. This is the most flexible way to get text out of the models in a new scenario.

When to use it
a new support ticket arrives and you want a first draft ready in the helpdesk before the agent opens it.
Watch out
the stored response stays on OpenAI until you remove it with Delete a response.
OpenAI (ChatGPT, Whisper, DALL-E)19

Generate a transcription

Action

In MakeTranscribes an audio to text.

Turns an audio recording into written text you can search, quote and edit. Voice notes, calls and interviews stop being a black box.

When to use it
a podcast episode or a sales call is saved to your drive and you want the transcript next to it for show notes or CRM notes.
Watch out
on the Free plan, files over 5 MB are refused by Make itself, so long recordings call for a paid plan.
OpenAI (ChatGPT, Whisper, DALL-E)20

Generate a translation (Whisper)

Action

In MakeTranslates an audio to English.

Listens to an audio file and returns English text, whatever language was spoken. It transcribes and translates in the same step.

When to use it
your team in the US receives voice messages from partners in Spain or Brazil and wants to read them straight away in English.
Watch out
English is the only output language; to keep the original language, use Generate a transcription instead.
OpenAI (ChatGPT, Whisper, DALL-E)21

Generate a video

Action

In MakeGenerates a video using Sora models with a text prompt and optional image or video references.

Asks a Sora model for a video from a text prompt, optionally guided by a reference image or clip. The module starts the job; the video itself follows.

When to use it
a new product row in your sheet should come with a short teaser video for social media, described by the marketer in one line.
Watch out
treat the result as a job and check it with Get a video before sending anything to a publishing tool.
OpenAI (ChatGPT, Whisper, DALL-E)22

Generate images

Action

In MakeGenerates images using GPT Image.

Creates pictures from a written description with GPT Image. You describe the scene, you get image files back to store or publish.

When to use it
every new blog draft in your editorial board needs a header illustration, and the writer already typed a short description of it.
Watch out
if you already have a photo to adjust, Edit images keeps the original and changes only what you ask.
OpenAI (ChatGPT, Whisper, DALL-E)23

Generate speech from text

Action

In MakeGenerates an audio file based on text input and settings.

Reads a text out loud and delivers an audio file, using the settings you choose in the module. Articles, instructions or messages become something people can listen to.

When to use it
a training platform publishes a new lesson and you want an audio version attached for learners on the move.
Watch out
the output is a file, so send it to a module that stores or attaches files rather than to a text field.
OpenAI (ChatGPT, Whisper, DALL-E)24

Get a batch

Action

In MakeRetrieves details of the specified batch.

Returns the details of one batch: where it stands and what it produced. It answers the question of whether your large job is done yet.

When to use it
right after Watch batch completed, to read the batch's details before sending its results to your catalog or data warehouse.
Watch out
map the batch from the previous module instead of pasting it by hand, so the scenario works for every batch and not just the one you tested.
OpenAI (ChatGPT, Whisper, DALL-E)25

Get a response

Action

In MakeRetrieves an existing model response.

Fetches a model response that already exists, so a later step or another scenario can reuse the text without asking the model again.

When to use it
a draft was generated when the ticket arrived, and a second scenario picks it up once a human approves it for sending.
Watch out
it only works as long as the response is stored; after Delete a response, this module has nothing left to return.
OpenAI (ChatGPT, Whisper, DALL-E)26

Get a skill

Action

In MakeReturns information about a specific skill by its ID.

Reads the information of one skill from its ID. You see what the skill is before deciding what to do with it next.

When to use it
a request in your project tool mentions a skill, and you want its details added to the request so the reviewer has context.
Watch out
this returns information about the skill, not the package; for the files themselves, Download a skill is the right module.
OpenAI (ChatGPT, Whisper, DALL-E)27

Get a skill version

Action

In MakeReturns information about a specific skill version.

Returns what is known about one precise version of a skill. It is the check you run before archiving, promoting or deleting that version.

When to use it
before a new version becomes the default, you want its details posted to the approver so the change is visible.
Watch out
keep the skill and the version apart when mapping; mixing the two IDs is the classic error with this family.
OpenAI (ChatGPT, Whisper, DALL-E)28

Get a video

Action

In MakeReturns information about a specific video by its ID.

Pulls up the information on one video from its ID, including where the job stands. It is the natural follow-up to any module that creates or lists videos.

When to use it
Watch video jobs spots a new job and you want its details before posting a message to the content channel.
Watch out
a video just created may not be ready yet, so read its status before sharing it.
OpenAI (ChatGPT, Whisper, DALL-E)29

Make an API call

Action

In MakePerforms an arbitrary authorized API call.

Sends any authorized request to the OpenAI API with the connection you already set up. It covers every endpoint that the listed modules leave out.

When to use it
OpenAI releases a feature that has no module in Make yet, and you want to use it in your scenario today.
Watch out
this is the one module where you describe the request yourself, so keep the API reference open while you fill it and test with Run once.
OpenAI (ChatGPT, Whisper, DALL-E)30

Message an assistant

Action

In MakeSend messages to a specified or newly created thread and execute it seamlessly. This action can send the arguments for your function calls to the specified URLs (POST HTTP method only). Works with Assistants v2.

Sends messages to an assistant thread, new or existing, and runs it. When the assistant decides to call one of your functions, the module can pass the arguments to a URL you choose, by POST only. It works with Assistants v2.

When to use it
a customer question arrives by email and your assistant should look up the order in your system before answering.
Watch out
the receiving URL must accept POST.
OpenAI (ChatGPT, Whisper, DALL-E)31

Remix a video

Action

In MakeGenerates a video remix using an existing video and text prompt.

Starts from a video you already generated and a new prompt, and produces a variation of it. You keep what worked and change the rest.

When to use it
a teaser was approved, and the regional team wants the same clip with a different setting for their market.
Watch out
like Generate a video, it gives you a job first, so read the result with Get a video before publishing.
OpenAI (ChatGPT, Whisper, DALL-E)32

Simple text prompt

Action

In MakeGenerates text responses from prompts without requiring an account or connection.

Asks a question and returns a text answer, with no OpenAI account and no connection required. It is the quickest way to try AI in a scenario.

When to use it
you want to test whether a summary or a rewrite helps your workflow before creating an OpenAI account for the team.
Watch out
once the test convinces you, move to Generate a response, which runs on your own connection.
OpenAI (ChatGPT, Whisper, DALL-E)33

Transform text to structured data

Action

In MakeIdentifies information in a prompt's text and returns it as structured data.

Reads free text, spots the pieces of information you asked for, and returns them as separate values. A messy email becomes clean fields ready to map.

When to use it
quote requests arrive as plain emails, and you want the company, the budget and the deadline sent as separate columns to your CRM.
Watch out
name every piece of information you expect; what you do not ask for will not come back.
OpenAI (ChatGPT, Whisper, DALL-E)34

Update the default version pointer for a skill

Action

In MakeUpdates the default version pointer for a skill.

Changes which version of a skill is used by default. The versions stay as they are; only the pointer moves, so rolling back is as easy as pointing to the older one.

When to use it
a new version passed review and should go live, or the latest one misbehaves and you want the previous one back.
Watch out
Create a skill version alone never changes what is live; this module does.
OpenAI (ChatGPT, Whisper, DALL-E)35

Upload a file

Action

In MakeUploads a file to be used across the OpenAI platform.

Sends a file to OpenAI so other features of the platform can use it later: vector stores, batches, assistants. It is the entry door for your documents.

When to use it
a contract, a price list or a batch input file lands in your drive and needs to be available to OpenAI.
Watch out
on the Free plan Make caps files at 5 MB, so large files need a paid plan.
OpenAI (ChatGPT, Whisper, DALL-E)36

List batches

Search

In MakeRetrieves a list of batches.

Returns the batches of your account, each one as its own bundle. It gives you an overview of the big jobs in progress or finished.

When to use it
you want a status report of every batch copied into a sheet so the team sees at a glance what is still running.
Watch out
a search can return several bundles, and every module placed after it runs once per batch, which counts as more operations.
OpenAI (ChatGPT, Whisper, DALL-E)37

List input items

Search

In MakeLists input items.

Lists the input items behind a model response, meaning what was sent to the model to produce it. Handy to understand why an answer looks the way it does.

When to use it
a generated reply was flagged as off topic, and the reviewer wants to see exactly what the model received.
Watch out
expect several bundles, one per item, so add an aggregator if you want them in a single message.
OpenAI (ChatGPT, Whisper, DALL-E)38

List skills

Search

In MakeReturns a list of skills.

Returns every skill in your account, one bundle each. It is the starting point when you want to act on skills without knowing their IDs in advance.

When to use it
an audit sheet should show all skills in use, refreshed each time the scenario runs.
Watch out
never chain it directly into Delete a skill without a filter in between, or every skill in the list goes.
OpenAI (ChatGPT, Whisper, DALL-E)39

List skill versions

Search

In MakeReturns a list of skill versions.

Shows all versions of one skill, as separate bundles. You see the full history of a skill before picking the one to promote, download or remove.

When to use it
a reviewer needs to compare the current rules with earlier ones before signing off on a change.
Watch out
combine it with a filter to keep only the version you want, since the modules after it run for each version returned.
OpenAI (ChatGPT, Whisper, DALL-E)40

List video jobs

Search

In MakeReturns a list of video jobs.

Returns the video jobs of your account, one bundle per job. Useful for a view of everything Sora is making or has made for you.

When to use it
the content lead wants a recap sheet of every job with its status, rebuilt whenever the scenario runs.
Watch out
each job becomes a bundle, and downstream modules run for every one of them.
Need help

Need help automating OpenAI (ChatGPT, Whisper, DALL-E) with Make?

A person reads every message.

FAQ

Questions people ask next

01Is the OpenAI (ChatGPT, Whisper, DALL-E) Make integration free?
Yes, on the Make side. OpenAI is a standard app, available from the Free plan. That plan gives you 2 active scenarios, a minimum interval of 15 minutes between scheduled runs, 5 minutes of execution per run, files up to 5 MB and 512 MB of data transfer. Paid plans (Core, Pro, Teams, Enterprise) bring the interval down to 1 minute and lift the active scenario cap. Usage is counted in operations. OpenAI's own billing for the models is a separate matter that this page does not cover.
02What do you need to connect OpenAI to Make?
An OpenAI account and a Make account. Add any OpenAI module, click Create a connection, give it a name if you like, then authorize Make on OpenAI's page or paste the key OpenAI gives you, and click Save. The official documentation lists no other prerequisite. One exception is worth knowing: Simple text prompt answers without any account or connection, so you can try the integration before setting anything up. Every other module, triggers included, uses the connection.
03Do the OpenAI triggers fire instantly in Make?
No. Watch batch completed and Watch video jobs are both scheduled triggers, shown with a clock badge. Make checks OpenAI each time the scenario runs, every 15 minutes by default. On the Free plan the minimum gap is 15 minutes; on paid plans it goes down to 1 minute. There is no instant trigger in this app, so a finished batch or a new video job is picked up at the next check. If you need a scenario to start the moment something happens elsewhere, start it from the other app's trigger instead.
04What if the OpenAI module you need is missing in Make?
Use Make an API call. It reaches any endpoint of the OpenAI API with the connection you already have, so a feature without its own module is still within reach. The official documentation lists no specific limits for this app. For large volumes, Create a batch and Watch batch completed let OpenAI process many requests together. If a scenario fails, the Make troubleshooting guide on this site covers the common causes.
05Should you use Make or n8n for OpenAI?
It depends on where your other automations already live. OpenAI exists in both tools: Make offers 42 modules, and n8n has its own OpenAI node documented on a separate page of this site. Building in the tool your team already uses keeps connections and habits in one place. Compare the modules each side offers for your exact use case, such as transcription, images or batches, and pick the one that covers it without extra work. Neither page claims one is faster or cheaper.