An honest 2026 review of Captions AI (now Mirage): its mobile video editor, auto-captions, dubbing, and AI avatars — what it nails and where it stops.
Captions is one of the best mobile AI video editors you can buy: auto-captions in 100+ languages, text-based editing that strips filler and fixes eye contact, lip-synced dubbing, AI avatars you can spin from a selfie, and a Create Clips tool that can batch a long recording into dozens of ranked shorts. For finishing vertical video on a phone, it is excellent and earns its reputation. The honest ceiling is scope — it is a video-only editor, not a content operation. It makes no carousels, blogs, or newsletters, and has no publisher or scheduler, so everything you make leaves as an export you post by hand. Rate it 3.8 as a content tool: a superb maker that stops at the render.
Captions (the company rebranded to Mirage in September 2025 and later folded its Mirage Studio avatar product back into the app) is the video editor a lot of creators reach for first when they want captions and clean talking-head edits off their phone. It is made by a well-funded New York startup — co-founded in 2021 by Gaurav Misra and Dwight Churchill, backed by Sequoia, Andreessen Horowitz, Index Ventures and others, with more than $100 million raised — and it shows in the polish.
This review scores it on the dimensions that matter to a creator: the editing, captions, dubbing, avatars, and clip generation, where it is genuinely strong, plus multi-format breadth, brand governance, and publishing, where a video-only mobile editor necessarily scores low. Both halves are real. The high scores are earned; the low ones are about what the product was never built to do, not a knock on its craft.
Everything here reflects the app's state as of 2026-10-01. Captions has changed its plan names and prices more than once across 2025–2026, and feature placement shifts with releases, so treat pricing and the exact menu of tools as a moving snapshot and confirm them in the app. Where a figure could drift, I generalize rather than state false precision.
Captions is an AI video creation and editing app for iOS, Android, and web. Its core loop is "upload footage, skip post-production": it auto-generates captions (with customizable fonts and styles in 100+ languages), cuts scenes, overlays B-roll, removes filler words and awkward pauses, corrects eye contact, cleans up audio, and can add AI music and sound effects — much of it driven by plain-language, text-based editing prompts rather than a manual timeline. Its Create Clips tool takes a long recording (a podcast or webinar, up to 3 hours or 10GB) and auto-generates up to 30+ clips, each reframed and ranked by an AI virality score. Its dubbing tool (originally Lipdub) translates a video into dozens of languages with synced lip movement. The other half of the product is generative: AI avatars and digital twins. You can create an AI actor from a selfie or a short clip and reuse that persona across new scripts, outfits, and backgrounds, or generate a talking "AI Creator" video from text — the capability that came out of Mirage Studio and is now merged into the main app. What Captions does not include is the layer around a finished video: there is no carousel, blog, or newsletter generation, and no scheduler or multi-platform publisher. Every output — one edit or a batch of clips — is a video file you export and post yourself.
Captions fits the solo creator, UGC maker, or small social team whose bottleneck is finishing short vertical video on mobile — adding captions, tightening a talking-head cut, dubbing into another language, clipping a long recording into shorts, or standing up a spokesperson avatar without a shoot. For that person it is fast, polished, and often the single best tool for the job. It fits poorly for anyone who needs a content operation rather than video: a creator running a multi-format calendar (video and carousels and a blog and a newsletter), holding one brand voice across dozens of posts, and publishing on a schedule across platforms will find Captions does only the video step — however much video — and leaves the rest to other tools.
| Dimension | Score | Why |
|---|---|---|
| AI video editing | 4.5 / 5 | Text-based editing, auto-cuts, B-roll overlay, filler and pause removal, and eye-contact correction genuinely remove post-production grunt work. |
| Auto-captions & styling | 4.6 / 5 | Its namesake feature: fast, accurate captions with rich styling in 100+ languages — among the best in the category. |
| Dubbing & translation | 4.3 / 5 | Lip-synced dubbing across dozens of languages is strong for localization, though quality varies by language. |
| AI avatars & digital twins | 4.1 / 5 | Create an AI actor from a selfie and reuse it across scripts; genuinely useful, but credit-hungry and the uncanny tell shows on longer scripts. |
| Mobile experience & ease of use | 4.5 / 5 | Mobile-first and polished — you can go from raw footage to a finished vertical video entirely on a phone. |
| Clipping long-form at scale | 4.0 / 5 | Create Clips can turn a long podcast or webinar (up to 3 hours) into 30+ reframed, AI-ranked shorts in one pass — a genuine strength, not a gap. |
| Multi-format breadth | 1.5 / 5 | Video only — no carousels, blog articles, or email newsletters, so it covers one of the formats a full calendar needs. |
| Brand voice & persona governance | 2.0 / 5 | Edit styles and reusable twins help, but there is no cross-format brand-voice system or banned-word governance across a batch of posts. |
| Publishing & scheduling | 1.0 / 5 | None. Every output is an export you post by hand; there is no scheduler, no queue, and no multi-platform publisher. |
| Value & pricing | 3.3 / 5 | The credit system rations AI video and avatar generation; heavy use pushes you into the expensive Frontier tiers, and prices are iOS-anchored. |
Captions runs on a credit-based subscription, with prices shown for iOS and discounted annual billing available. There is a free app for basic editing and captioning (watermarked exports), a mid tier around $24.99/mo that unlocks the curated AI edit styles, digital-twin creation, AI B-roll, music and images with a monthly credit allowance, and higher "Frontier" tiers that add a more capable set of generative models with larger credit pools, scaling up toward roughly $280/mo at the top before enterprise. Treat the exact numbers as a moving target — Captions has reshuffled plan names and prices repeatedly.
For what it is — a mobile AI editor plus avatar generation — the pricing is fair, and the free tier is a real way to try it. The catch is the credit math once avatars and AI video enter the picture: generative minutes burn credits quickly, and a creator producing volume can blow through a mid-tier allowance and get pushed up the Frontier ladder. The value question is less "is Captions overpriced" and more "how much of my content does this one tool actually cover."
That is where the comparison to a content engine matters. Captions prices the making of a video; it does nothing for the clips, carousels, captions-on-everything, blog, newsletter, and scheduling a real content presence needs around it — that stays a separate stack. Against an engine that bundles generation and publishing (Kompozy runs Starter at $199/mo and Pro at $499/mo, generates 18 formats, and publishes across nine platforms), Captions is the better way to finish one standout video and the more expensive way — in time and bolt-on tools — to run an ongoing, multi-platform operation. They price for different jobs.
| Use case | Fit | Why |
|---|---|---|
| Adding styled auto-captions to a vertical video | Strong | This is the product's core strength — fast, accurate, well-styled captions in many languages. |
| Editing a talking-head clip quickly on mobile | Strong | Text-based editing, filler removal, and eye-contact correction make phone-only finishing genuinely fast. |
| Dubbing a video into other languages | Strong | Lip-synced dubbing across dozens of languages covers the localization job well. |
| Making a spokesperson/UGC avatar without filming | OK | Digital twins and AI actors work, but they are credit-hungry and realism varies on longer scripts. |
| Turning one long podcast into a week of clips | Strong | Create Clips mines a long recording for up to 30+ ranked, reframed shorts in one pass — this is a real strength, not a gap. |
| Running a multi-format calendar (video + carousels + blog + email) | Weak | It makes video only — no carousels, blog articles, or newsletters. |
| Publishing and scheduling across platforms | Weak | There is no publisher or scheduler; every output is a manual export and upload. |
| Holding one brand voice across dozens of posts | Weak | Edit styles and twins help per video, but there is no cross-format brand-voice or banned-word governance. |
It would be a category error to score Captions against Kompozy head-to-head — one finishes video (one clip or a batch of them), the other runs the operation around it. This review deliberately rates Captions high on editing, captions, dubbing, avatars, and clipping at scale, where it is excellent, and low on multi-format breadth and publishing, where it does nothing, because that split is the whole point. A beautifully captioned, dubbed, eye-contact-corrected batch of vertical clips is still video, sitting in your camera roll with nowhere to go.
Kompozy is where that distance closes. Take the finished Captions export and Kompozy clips it into more shorts, reframes it for each feed, and spins the same source into a brand-exact [Carousel](/glossary/hyperframes), [Quote Graphics](/glossary/output-buckets) of the best lines, a Blog Article, and an Email Newsletter — all in one voice held by a [Persona Brief](/glossary/persona-brief) — then schedules and publishes the whole set across the eight social platforms plus blog and email on [Autopilot](/glossary/autopilot) with a review step. It also generates net-new formats Captions does not, including face-locked [Persona Shorts](/glossary/persona-shorts) from its own avatar pool. The honest reading: Captions is a best-in-class way to make the hero video; Kompozy is how that video — and everything it could become — actually ships. Many creators run both.
For finishing short vertical video on a phone — captions, quick talking-head edits, dubbing, batch-clipping a long recording, and spokesperson avatars — yes, it is one of the best tools available and the free tier lets you try it first. It is worth less if your need is a full content operation: everything it makes is still video, with no carousels or blog or newsletter, and no publisher or scheduler.
It uses a credit-based subscription with iOS-shown prices and discounted annual billing. There is a free tier with watermarked exports, a mid tier around $24.99/mo, and higher "Frontier" tiers with larger credit pools that scale up toward roughly $280/mo before enterprise. Captions has changed plan names and prices more than once, so confirm the current figures in the app.
Effectively yes. The company rebranded to Mirage in September 2025 to signal a pivot toward AI video research, and it later merged its Mirage Studio avatar product back into the main app — which is still called Captions on captions.ai. So Mirage is the company/model side, and Captions is the app you use.
No. Captions finishes a video and you export it; there is no built-in scheduler and no multi-platform publisher. Posting to TikTok, Reels, Shorts, LinkedIn, and the rest is a manual step. If you want generation plus scheduled publishing in one place, that is the job of a content engine like Kompozy.
They overlap heavily on captions and short-form editing. Captions leans harder into AI avatars, digital twins, dubbing, and batch-clipping a long recording; Submagic is tightly focused on fast captioned short-form. For avatar/spokesperson work and clipping at scale Captions usually wins; for pure caption-and-cut speed it is close. Neither publishes across platforms.
Yes — its Create Clips tool takes a long recording (a podcast or webinar, up to 3 hours or 10GB) and auto-generates up to 30+ clips, reframed and ranked by an AI virality score. What it doesn't do is anything past the video: no carousels, blog, or newsletter from the same source, and no publisher to get the clips onto a platform.
For captions and short-form editing, Submagic and CapCut are the closest. For a dedicated long-to-short clipper, OpusClip. For transcript-based editing and dubbing, Descript. And for turning a finished video into a full multi-format, multi-platform content stream, Kompozy is the engine that sits around whatever editor you use.