Freepik launched Spaces in November 2025, adding a node-based AI canvas to its already massive creative ecosystem. For teams that live inside Freepik's stock library, it felt like a natural extension. But if you need deep model access, 3D generation, or Gaussian splatting, the story gets more complicated.
I've been testing both Freepik Spaces and Raelume over the past few weeks. Here's how they compare across the things that actually matter for creative workflows.
Quick Comparison
| Feature | Freepik Spaces | Raelume |
|---|---|---|
| Canvas Type | Node-based, infinite | Node-based, infinite |
| AI Models | 36+ image models, 9+ video models, audio + editing tools | 70+ models across all media types |
| Media Types | Image, Video, Audio | Image, Video, 3D, Audio, Text, WORLDS |
| 3D / Gaussian Splatting | No | Yes (WORLDS blocks) |
| Real-time Collaboration | Yes | Yes (unlimited team members) |
| Stock Library | Massive (Freepik ecosystem) | No built-in stock library |
| Pricing Model | Credit-based tiers ($5.75 to $158/mo) | Free tier with credits, paid plans |
| Free Tier | Up to 3 Spaces, limited credits | Free credits, no credit card required |
The Canvas Experience
Both tools use a node-based approach where you connect blocks visually. If you've used ComfyUI or any visual programming tool, the paradigm will feel familiar.
Freepik Spaces: connecting image generators in a node workflow to composite characters into environments.
Freepik Spaces integrates tightly with the Freepik stock library. You can pull in vectors, photos, and templates directly from their catalog, which is genuinely useful if your team already relies on Freepik assets. The collaboration features work well, with Figma-style cursors and shared editing.
Raelume's canvas: each block has inputs and outputs connected by edges, with content flowing between generation steps.
Raelume takes a different approach to scope. Rather than anchoring to a stock library, it focuses on giving you access to as many AI models as possible across as many media types as possible. The block system follows the same input/output logic, but the range of what you can connect is significantly wider.
Model Access: Where the Gap Gets Real
Freepik has quietly built one of the largest model libraries in the space. On the image side alone, they offer 36+ models including Flux (with Kontext for text-in-image), Mystic (their own model built on Flux), Google Imagen, Ideogram, Seedream, Nano Banana Pro, Magnific for upscaling, Runway, GPT/DALL-E, and Z-Image Turbo. For video, the lineup is equally deep: Google Veo 3, Kling 2.1/2.6/3.0, Runway Gen 4, Seedance, Wan AI, PixVerse 4.5, MiniMax Hailuo 02, and Sora. Audio gets ElevenLabs, Sound Effects generation, and a Lip Sync API. On top of that, there's a full suite of editing tools: image upscaler (up to 10K resolution), video upscaler, background remover, retouch, reimagine (image-to-image variations), image enhancer, sketch-to-image, and a video editor.
That's a serious toolkit. For pure image and video generation with built-in editing, Freepik is hard to underestimate.
Raelume takes a different approach with 70+ models. That includes Flux 2 Pro Ultra, Nano Banana Pro (4K output), Kling 3 Pro, Veo 3.1 (4K video), ElevenLabs V3, Hunyuan3D v3 for 3D, and Claude Opus 4.6 for text. Both platforms share several popular models (Flux, Kling, Nano Banana Pro, ElevenLabs). Where Raelume pulls ahead is media type breadth: image, video, 3D modeling, audio, text generation, and WORLDS (more on that below). Freepik covers image, video, and audio exceptionally well, plus editing. But it doesn't touch 3D generation or spatial content.
The WORLDS Feature: Something Nobody Else Has
Raelume supports generation across image, video, 3D, audio, text, and WORLDS blocks.
Raelume's WORLDS blocks do something I haven't seen in any competing canvas tool: Gaussian splatting. You can take a 2D image, convert it into a 3D environment using Gaussian splatting, add 3D objects to the scene, move a virtual camera freely, and capture 2K to 4K images from any angle.
This is genuinely novel. Neither Freepik Spaces, nor Krea (which has its own impressive node-based canvas with 50+ models), nor any other tool in this category offers Gaussian splatting as part of the creative workflow. For teams working on spatial content, VR previsualization, or 3D asset pipelines, this is a meaningful advantage.
Pricing: Credits vs. Simplicity
Freepik's pricing is where things get confusing. The tiers look reasonable at first glance: Free (limited), Essential at $5.75/month (84K credits per year), Premium at $12/month (216K credits), Premium+ at $24.50/month (540K credits), and Pro at $158.33/month (3.6M credits).
The problem is credit costs vary wildly. A single image generation can cost anywhere from 50 to 500 credits depending on the model and settings. A 9-second HD video eats roughly 2,600 credits. "Unlimited" image generation only kicks in at the Premium+ tier, and even then, advanced features still consume credits. For teams doing heavy video or multi-model work, it's difficult to predict monthly costs.
Raelume offers a free tier with no credit card required and free credits to start. The paid plans scale for teams. The pricing structure is simpler, though both platforms ultimately charge based on usage at higher volumes.
Where Freepik Wins
Let's be fair. Freepik has real advantages:
Massive model and editing toolkit. 36+ image models, 9+ video models, and a full suite of editing tools (upscaler up to 10K, background remover, retouch, sketch-to-image, video editor). For teams that need generation plus post-processing in one place, this is comprehensive.
Stock library integration. If your workflow involves pulling stock photos, vectors, or templates, Freepik's native integration is hard to beat. Raelume doesn't have a built-in stock library.
Brand ecosystem. Freepik is a household name in creative tools. The Spaces canvas plugs into an ecosystem that includes Freepik's image editor, mockup tools, and massive asset catalog. For teams already paying for Freepik, Spaces is a natural add-on.
Lower entry price. The Essential plan at $5.75/month is an accessible starting point, even if credits run out faster than you'd expect.
Where Raelume Wins
Model depth. 70+ models across six media types versus a curated handful. For teams that need flexibility, this matters.
3D and WORLDS. Gaussian splatting and 3D generation are simply not available in Freepik Spaces. If your pipeline touches 3D at all, this is the deciding factor.
Collaboration scale. Raelume offers unlimited team members. Freepik's collaboration features are solid but the free tier caps you at three Spaces.
Transparent usage. No confusing credit tiers where the same action costs different amounts depending on which model you pick.
Who Should Use What
Choose Freepik Spaces if: you already live in the Freepik ecosystem, your work centers on 2D image and video content, and you want tight stock library integration. The credit system takes getting used to, but if your volume is moderate, the lower tiers offer decent value.
Choose Raelume if: you need access to a wide range of AI models across multiple media types, your workflow includes 3D or spatial content, or you want to avoid juggling credits and subscriptions across different AI tools. The WORLDS feature and 70+ model library put it in a different category for teams doing complex, multi-format creative work.
The Bigger Picture
The AI canvas space is moving fast. Krea is doing impressive things with real-time generation. Fuser is pushing 200+ models. Freepik is leveraging its massive user base to bring AI workflows to a broader audience.
What's interesting about this moment is that each tool is making a different bet. Freepik bets on ecosystem integration. Raelume bets on model breadth and novel capabilities like Gaussian splatting. Krea bets on real-time interaction.
For creative teams evaluating their options, the real question isn't which tool is "best." It's which tradeoffs align with how you actually work. Stock access or model variety? 2D focus or multi-format pipelines? Familiar ecosystem or cutting-edge features?
The answer depends on the work.
Alex Mercer reviews AI creative tools as an independent writer. No affiliations, no sponsorships.
Top comments (0)