Vimpo vs Immersive Translate: no subscription, your own model

Immersive Translate is how a lot of people follow foreign-language YouTube, and it's good at it. Vimpo gives you bilingual subtitles too, so people ask what the difference is.

Three things, mostly: whether you pay monthly, whose model does the translating, and what you can make from the video afterwards. Same video, same minute:

In VimpoIn Immersive Translate

Same video, same minute

Andrej Karpathy — We’re summoning ghosts, not building animals

¿Qué crees que va a tardar una década en lograrse?

What do you think will take a decade to accomplish?

1:46 / 2:26:08Both

No subscription, your own model

Immersive Translate's free plan has a daily limit, and the better models come with Pro, which is a monthly subscription. Vimpo is a one-time purchase, and translation runs on your own API key, from DeepSeek, OpenAI, Claude, Gemini and others. You pay the provider directly for what you use, with nobody in between. Or run a local model through Ollama or LM Studio and pay nothing. Speech recognition runs on your Mac, with no quota.

Every batch carries the title and the whole transcript, so names and terms stay consistent. Same model, with that off and on:

Make something from it

Subtitles are where Vimpo starts. In Studio (⌘2) you pick lines on the transcript and press ⌘B to make a clip, AI can suggest the parts that stand on their own, and every clip exports with its subtitles burned in. Need them in another editor? Export SRT or ASS.

1. 01

   Pick lines in the transcript

   Click a line, then Shift-click another.

2. 02

   Press ⌘B to make a clip

   Trim it on the timeline; its edges snap to the lines.

3. 03

   Let AI find highlights

   Your own model reads the subtitles and picks moments that stand on their own.

4. 04

   Style the subtitles

   Outline, box or large type, in one language or two.

5. 05

   Export

   Subtitles burned in, or FCPXML to keep editing in Final Cut or Premiere.

Click a line, then Shift-click another.

Trim it on the timeline; its edges snap to the lines.

Your own model reads the subtitles and picks moments that stand on their own.

Outline, box or large type, in one language or two.

Subtitles burned in, or FCPXML to keep editing in Final Cut or Premiere.

Andrej Karpathy — We’re summoning ghosts, not building animals

CinemaStudio

Export

Transcript

1:21En realidad, es mucho más exacto llamarla la década de los agentes.This is really a lot more accurately described as the decade of agents.

1:46¿Qué crees que va a tardar una década en lograrse?What do you think will take a decade to accomplish?

1:56Piénsalo casi como un empleado o un becario que contratas para trabajar contigo.You should think of it almost like an employee or an intern that you would hire to work with you.

2:13Si hoy no lo haces, es porque todavía no funcionan.The reason you don't do it today is because they just don't work.

2:23No tienen aprendizaje continuo. Les cuentas algo y no lo recuerdan.They don't have continual learning. You can't just tell them something and they'll remember it.

2:28Resolver todos esos problemas llevará más o menos una década.It will take about a decade to work through all of those issues.

2 lines selected

1:46 – 2:13 · 0:27

Make Clip

En realidad, es mucho más exacto llamarla la década de los agentes.

This is really a lot more accurately described as the decade of agents.

¿Qué crees que va a tardar una década en lograrse?

What do you think will take a decade to accomplish?

Si hoy no lo haces, es porque todavía no funcionan.

The reason you don't do it today is because they just don't work.

Si hoy no lo haces, es porque todavía no funcionan.

The reason you don't do it today is because they just don't work.

00:01:46:12

IO

ClipsSubtitle Style

No clips yet

Clips1 · 0:27 total2 · 0:52 total3 · 1:15 total

The decade of agents1:21 – 1:46 · 0:25

What will take a decade?1:46 – 2:13 · 0:27

Why agents don't work yet2:13 – 2:36 · 0:23

AI Highlights

Moments from the subtitles that stand on their own

Find Highlights

AI Picks2 · 0:48 totalAdd All

1The decade of agents1:21 – 1:46 · 0:25

Add to Clips

2Why agents don't work yet2:13 – 2:36 · 0:23

Presets

SubtitlesOutlineSubtitlesBoxSubtitlesLarge

Two Languages

Translation on TopOriginal on Top

TimelineAI Picks 2

PictureSpanishSound

1:301:452:002:152:30

En realidad, es mucho más exacto llamarla la década de los agentes.¿Qué crees que va a tardar una década en lograrse?Piénsalo casi como un empleado o un becario que contratas para trabajar contigo.Si hoy no lo haces, es porque todavía no funcionan.No tienen aprendizaje continuo. Les cuentas algo y no lo recuerdan.Resolver todos esos problemas llevará más o menos una década.

1

2

The decade of agents

What will take a decade?

Why agents don't work yet

Export

The decade of agents0:25

What will take a decade?0:27

Why agents don't work yet0:23

Subtitles

Burn InSRT File

Quality

720p1080p

FCPXML, to keep editing in Final Cut / Premiere

CancelExport 3 Clips…

For watching as you browse, Immersive Translate is easier

There's nothing to download: open the page and the subtitles are there, and it can generate them during playback for videos that have none. It works on Netflix and other protected sites Vimpo can't download from. It also translates web pages, PDFs and ebooks, and it runs on your phone.

Side by side

The Immersive Translate column is based on its website and membership terms (October 2026). In that column, a dash means its site doesn't mention the feature.

Which one to use

• You'd rather not add another subscription, and want your own API key or a local model: Vimpo.
• You make clips with subtitles from long videos: Vimpo.
• Following YouTube or Netflix as you browse, plus web pages and PDFs: Immersive Translate.

They work well together: browse with Immersive Translate, and paste the ones worth cutting into Vimpo. Every Vimpo feature is free during the beta. Download it here.