A company casually gives up a brand it spent 14 years establishing around the world in favor of the name of a startup it acquired only two years earlier. That reversal happened in spring 2026, in a market where generative AI is becoming central to creative production.
This episode offers a remarkably vivid picture of what is changing in the work of “making.” Magnific became a symbol of that change, while Higgsfield AI is pursuing the same market with a completely different philosophy. Placing these two platforms and their contrasting design principles side by side reveals the outline of the next generation of production environments.
The Acquirer Took the Acquired Company’s Name
Tracing the roots of one of the main players, Magnific , leads to a long-established Spanish company. Founded in Málaga in 2010, Freepik grew for nearly 15 years without outside capital as a vast “asset warehouse” distributing vectors and stock photos.
The turning point came in 2022, when OpenAI released DALL·E 2. Quickly recognizing the arrival of an era in which images could be created on the spot, the company pivoted decisively toward generative AI. In May 2024, it acquired Magnific, a Spanish image-enhancement tool that had attracted 30,000 users within just 24 hours of launch. Its founders, including Javi López, remained with the business, and its momentum continued after the acquisition.
Then, on April 28, 2026, Freepik integrated both its corporate identity and services into Magnific, transforming itself from a “warehouse” that supplied assets into a “workshop” where creators make and finish them on the spot. The acquirer adopted the name of the acquired company. That alone shows how deeply the Magnific name had taken hold among professional creators.
The figures underscore the significance of this decision. The company reportedly became profitable without accepting any venture funding and reached approximately $230 million in annual recurring revenue. It is said to have more than one million paid subscribers and to be used in professional work by over 250 companies, including production teams at the BBC, Puma, and Amazon Prime Video. Several million images are generated each day, and a library of more than 250 million assets was carried forward under the new banner. Investor rankings have listed it among Europe’s leading AI creative companies.
A Film Studio Built Inside the Browser
The other platform, Higgsfield AI , is a video startup founded in the United States in 2023. Established by Snap’s former head of generative AI, it brings planning, casting, shooting, editing, and ad optimization together in a single browser-based workspace.
Its aim differs completely from that of a one-off tool that simply “turns text into a short video.” The concept is more like building a virtual film studio inside the browser.
Its greatest distinction is that it does not insist on using only its own models. OpenAI’s Sora 2, Google’s Veo 3.1, Kuaishou’s Kling 3.0, ByteDance’s Seedance 2.0, and Higgsfield’s own photorealistic DoP model can all be called up and used as needed.
This is combined with camera control resembling a film set. More than 50 professional compositions and moves—including “bullet time,” which circles a subject; a “crash zoom,” which rushes inward; and a “dolly zoom,” in which only the background appears to close in—can be selected with one click, elevating the user from an “AI operator” to a “cinematographer.”
A longstanding problem in sequential video has been that a character’s face, body shape, or clothing changes from shot to shot. Higgsfield addresses this with Soul Cast and Soul ID, systems that lock in a character by defining details such as genre, historical setting, personality type, build, and costume in advance. Once created, the person retains the appearance of the same individual even when placed in a different location, filmed with another camera, or processed through another model.
Marketing Studio can also automatically assemble several kinds of vertical short-form ads—such as unboxing videos, reviews, and tutorials—simply by receiving a product-page URL. Canvas, its node-based work surface, connects these stages into a workflow completed on one screen. The entire design relentlessly reduces the cost of moving from “I want to make this” to a finished result.
“Direction” or “Refinement”: This Is Where the Paths Divide
The two may look similar, but their underlying ideas point in opposite directions.
Higgsfield takes a top-down, direction-led approach. The creator first defines a character’s appearance, specifies the lens and camera movement, establishes the context of the entire scene, and then shoots the video from scratch. Most of its power is devoted to how firmly the creator can control intent at the entrance to production and how the story is constructed.
Magnific, by contrast, takes a bottom-up refinement approach. It starts with an existing photograph, rough sketch, or coarse asset created in another tool, then layers retouching and enhancement to push the final texture and resolution as far as possible.
Its signature “creative upscaling” does more than mathematically increase the pixel count. AI interprets and adds details that did not exist in the original data, down to skin texture, the weave of fabric, and reflections on metal. Controls for creativity, HDR, and fidelity to the source image allow users to adjust precisely how strongly AI intervenes. It also supports 4K video enhancement and is valued across the industry as a final-stage tool for turning rough generative-AI footage into material suitable for advertising and large screens.
Their natural audiences differ as well. Higgsfield suits marketers and D2C operators who want to mass-produce ads quickly from URLs, and video directors planning consistent long-form work starring AI characters. Magnific serves designers and art directors who want to build from existing brand assets or hand-drawn sketches and pursue final texture and resolution without compromise.
It is reportedly common in production to “complete 90% of the work in Higgsfield and entrust only the final finish before delivery to Magnific.” The platforms compete, but they can also act as partners with different roles.
Evaluate Pricing by Usage, Not Merely the Low Cost of Entry
Both companies combine a fixed subscription with credits that are consumed according to the processing demands of each generation. Higgsfield offers a free tier, while paid entry plans start at around $15 per month and higher tiers range up to approximately $99 per month with annual billing.
Magnific’s entry plan begins at just under $15 per month with annual billing, followed by plans for heavy users and shared team use.
The point to watch is that the actual amount paid varies far more with “which model you try, and how many times” than with the headline monthly fee. Premium models such as Sora 2 and Veo 3.1 consume many credits per generation, and repeated trials in pursuit of an ideal image may use more than expected. Prices and credit allocations are frequently revised, so checking the latest plans before subscribing is the safest approach.
The Real Contest Is Orchestration, Not Models
The paths of the two companies also illuminate a deeper market shift. The foundation models that generate images and video are rapidly becoming commonplace components. Neither Higgsfield nor Magnific chose to develop everything in-house and compete solely on raw generation quality. Instead, they chose to assemble excellent models from around the world and control them through a common interface.
In other words, the center of added value has moved from a technical contest over “which model produces the most beautiful image” to competition at the integration-and-control layer over “how generated assets fit into a real production workflow.” Freepik’s decision to discard the sign over its “warehouse” and become the Magnific “workshop,” and Higgsfield’s decision to place the Canvas work surface at its center, both follow precisely the same market logic. What users truly want is not a single powerful model but an “integrated work environment” that turns chaotic ideas into commercially finished products.
Cuenca’s View of a “No-Collar Economy”
Joaquín Cuenca, who leads Magnific, describes the meaning of this change as a “no-collar economy.” Just as the Industrial Revolution created blue-collar manual labor and the digital revolution produced white-collar office work, his view is that AI is creating an entirely new creative class constrained neither by physical labor nor by the old system of closed professional education.
Creating a brand video once required a camera operator, lighting technicians, costly equipment, and weeks of post-production. Today, a single marketer can paste one product URL and mass-produce videos resembling television commercials.
An individual can also give footage the resolution of a major studio from a home computer. In fact, roughly 70% of creators newly joining Magnific are reportedly “beginners,” suggesting that access to creative expression is indeed broadening.
Is your work centered on “direction and mass production from scratch,” or on “exhaustive refinement and ultra-high-resolution enhancement of existing assets”? Once that distinction is clear, the choice of Higgsfield or Magnific as the lead platform naturally follows. A level of expression once available only to organizations with enormous capital is moving into the hands of individuals. The two platforms are the twin engines supporting that tectonic shift.
