Mirage raises $75M to continue building models for its AI video editing app Captions
The startup behind Captions rebrands, trains a pacing model, and hits 200M videos created.
Mirage, the AI lab and company behind the popular Captions video editing app, has secured $75 million in growth financing from General Catalyst's Customer Value Fund. This major raise follows a strategic rebrand from 'Captions' to 'Mirage,' positioning the startup as an AI model developer for industries like advertising, not just a consumer app. Over the past year, the company has trained a proprietary AI model specifically for optimizing pacing, framing, and attention dynamics in short-form videos, a key differentiator in the crowded editing space. It also switched to a freemium model in January 2025 to compete with giants like CapCut and Meta's Edits.
The capital injection will accelerate Mirage's expansion into high-growth Asian markets and fund the development of new AI models focused on 'assembly intelligence'—automating the process of stitching together videos from various sources. The company has already launched a web-based marketing suite for bulk video creation, aiming to merge it with its mobile-first Captions app to better serve small businesses. CEO Gaurav Misra highlighted the success of their accent-preserving audio model, developed to serve its large international user base, which accounts for 75% of its revenue. With over 200 million videos created on its platform and strong unit economics praised by its investors, Mirage is betting its specialized, multi-model approach can win in the competitive AI video tool market.
- Raised $75M from General Catalyst to expand as an AI model lab beyond its Captions app.
- Trained a specialized AI model for video pacing/framing and launched a freemium web suite for marketers.
- The Captions app has 3.2M downloads, $28.4M in revenue, and 75% of income comes from outside the U.S.
Why It Matters
Signals a shift from single-app companies to vertically integrated AI labs building specialized models for professional use cases.