review
Magnific Review: The Platform We Generate Every Clip Through
Formerly Freepik, rebranded April 2026. We use Magnific to run our own benchmark, so read this one knowing we're not a neutral party, and we say so.
A disclosure before anything else: Magnific is the platform we generate every clip on this site through. We are not a neutral third party evaluating it from the outside, and you should read this review with that in mind. We have tried to write it the same way we would about any platform we have no relationship with, including reporting complaints we have not personally experienced.
What it actually is
Magnific was Freepik until 28 April 2026, when the whole company, stock library, AI tools, subscriptions and API keys included, rebranded under the Magnific name (Freepik had acquired the original Magnific AI upscaler in 2024 and folded it in). The company reports $230 million in annual revenue, over a million paying subscribers, and no outside VC funding; those are the company's own figures, not independently audited.
What it costs
Confirmed directly from Magnific's own pricing page. There is no free tier for individual plans; the entry point is a paid subscription:
| Plan | Price/mo | Price/mo annual | Credits/mo |
|---|---|---|---|
| Premium | €16 | €12 | 20,000, rolls over, no monthly reset |
| Premium+ | €36 | €27 | 45,000, plus unlimited Nano Banana 2 at 1K |
| Pro Starter | €88 | €66 | 112,500 or 300,000 (your choice), plus unlimited Nano Banana 2 at 2K |
API access draws from the same credit pool as the rest of your plan rather than being billed separately per call, which matters if you are budgeting a developer integration: there is no standalone per-call price sheet to check against, your subscription tier sets the ceiling.
The model catalog, and why we generate through here
Video: Seedance 2.0, Veo 3.1, Kling (2.5, 3.0 and the Omni variant), Runway Gen-4, Hailuo 02, PixVerse. Image: GPT Image 2, Nano Banana 2 and Pro, Imagen 4, Flux, Ideogram and more. That breadth, plus a documented API with an MCP server for direct integration into tools like Claude, is the specific reason we chose to run our benchmark generations through this platform rather than calling multiple labs' APIs separately. Every score on our leaderboard, including Kling 3.0 and Grok, was generated this way.
What we found when we looked past our own usage
We generate through Magnific regularly and have not personally hit the specific problems below, but we would be misrepresenting the platform if we only reported our own experience and ignored what other users report. Trustpilot reviews describe two-factor login codes arriving one to two hours late, past their expiry window, with templated support responses. A more serious pattern, if accurate: some users report that credit allowances that used to last around 20 days now exhaust in roughly a day since the rebrand, alongside reduced parallel-generation capacity and longer queue times. We have not independently benchmarked Magnific's own throughput or reliability the way we benchmark the models it gives us access to, so we cannot confirm or deny that pattern ourselves; we are reporting it because you should know it exists before you depend on this platform the way we do.
Should you use it
The model breadth and the developer-facing API are real strengths, and they are why this site runs on it. If you are evaluating it for a workload where consistent throughput matters, weigh the reliability complaints above against your own testing before committing, the same advice we would give for any platform, including this one.
We do not currently have a confirmed affiliate arrangement with Magnific; we are a paying customer, not a partner, which is its own kind of bias worth naming rather than an absence of one. See our full platform comparison and our affiliate disclosure for how we handle platforms we do earn a commission from.
Sources: Magnific, Magnific API docs, The Next Web, Trustpilot.
Related: Magnific
Priya Shenoy: Tracks what AI video actually costs across the platforms that resell access to the same handful of models. Treats a pricing page as a claim, not a fact, until someone checks it.