Queue 500 prompts.
Not 500 clicks.
DeepAI's own site generates one prompt at a time. Magic Rush queues hundreds, retries the ones that hit a rate limit, and files every result by project. Generation is included in your plan — there is no API key to buy and nothing to install.
Any modern browser · no API key needed · nothing to install
- donehero-product-shot-01.pngjob_8f21
- donehero-product-shot-02.pngjob_8f22
- runninglifestyle-cafe-loop.mp4job_8f23
- queuedbundle-mockup-04.pngjob_8f24
- queuedcampaign-still-09.pngjob_8f25
- retrysquare-variant-12.pngjob_8f26
Why this exists
A browser tab is not a job queue
Generating in bulk is mostly waiting. Doing it through a web page means the run is only as reliable as the tab it lives in.
The tab was the queue
Generation only continued while the page stayed open and focused. Switch tabs, sleep the laptop, close the window and the run was over — with no record of which prompts had already been sent.
A rate limit killed the batch
One rate-limit response partway through a 200-prompt run and everything after it failed, while everything before it had already been paid for.
Nobody could say what finished
Files landed in a downloads folder with no project, no order and no link back to the prompt that made them.
Side by side
The same model. A different job.
DeepAI makes the images. What it does not do is run four hundred of them while you are doing something else.
The provider's own page
- One prompt, one click, one wait
- Rate limits end the session
- Files land unnamed in your downloads
- No record of what you already made
- You watch the whole batch
Magic Rush
- Paste hundreds of prompts, one per line
- Rate limits are retried, not fatal
- Every result named and filed by project
- A queue that says what finished and what did not
- You leave the tab and come back
Same provider, same models, same output — the difference is the queue.
How it works
Four steps, once
- 01
Open the studio
No install and no account setup beyond your DeepAI key. Paste it once — it stays in this browser and is sent only to DeepAI.
- 02
Queue the batch
Paste or import your prompts, pick a project, set the concurrency and the delay between calls, and set a spend cap.
- 03
Close the tab
Leave the tab open and the queue works through the batch. A refresh resumes it; closing the tab ends it.
- 04
Collect the files
Outputs arrive saved and named by project. Anything that failed is listed with the reason, and can be requeued.
What it does
Built for the boring part
Not a prompt library and not a model. The work here is bookkeeping: knowing what was sent, what came back, and what to do when something fails.
- persistence
The queue survives a restart
Every job is saved as it changes state, so a refresh does not lose the run. The queue picks up where it stopped — including jobs that were mid-flight. Closing the tab does end the run; that is the one thing a browser page cannot do, and it is stated below rather than hidden.
- reliability
Retries instead of giving up
Rate limits and transient failures are retried with exponential backoff and jitter. A failure that never clears is reported as a failure, not quietly dropped.
- correctness
No duplicates, no silent losses
When a submission cannot be confirmed, the job says so. It is never re-sent automatically, because re-sending could spend your allowance twice, and it is never marked complete without an output to show for it.
- control
A spend cap you set
Set a ceiling inside your allowance and the queue stops when it is reached. A runaway batch on a large prompt list cannot quietly consume a whole month of generation overnight.
- organisation
Outputs filed by project
Every result is saved, named and sorted into its project, so a 500-file batch is a folder you can use rather than a downloads directory you have to sift.
- privacy
Generation is included
There is no API key to obtain and no separate bill. The site holds the provider credential and calls it on your behalf, against a monthly allowance set by your plan.
Pricing
One plan, generation included
Each plan includes a monthly allowance of images and video seconds. There is no API key to buy and no setup fee.
Solo
For one person generating on one machine.
$19/month
billed monthly · no setup fee · cancel any time
- Up to 3 active projects
- Queue of up to 500 jobs
- Persistent queue that survives a restart
- Automatic retry with backoff
- Monthly spend cap
- Organised, named downloads
Choose Solo- Most popular
Pro
For someone generating every day, across several projects.
$49/month
billed monthly · no setup fee · cancel any time
- Everything in Solo
- Unlimited projects
- Scheduled runs — start a queue at a set time
- Per-project spend caps
- Priority email support
Choose Pro Studio
For a small team working from a shared asset library.
$99/month
billed monthly · no setup fee · cancel any time
- Everything in Pro
- Up to 5 seats
- Shared project library
- Central spend controls across seats
- Owner dashboard covering every seat
Choose Studio
Every plan includes a monthly generation allowance. Cancel any time.
Questions
Before you subscribe
No. Generation is included in your plan. There is no DeepAI account to create and no second bill — the site holds the provider credential and meters each job against your plan's monthly allowance. You can see what is left on the studio page.
Stop babysitting a progress bar
Queue the batch, close the laptop, and come back to finished files that are already named and sorted.
Open the studio