AI Models & Disclosure
Last updated: October 29, 2026
reelmix is an AI-powered service operated by YC2123 LLC, a Wyoming limited liability company doing business as reelmix. We believe you should know exactly which AI technology creates your content. This page lists the AI models the Service currently uses, what each one does, and how we label and moderate AI-generated content.
reelmix is an independent product. We are not affiliated with, sponsored by, or endorsed by any of the model providers listed below. Model names and trademarks belong to their respective owners.
1. All Output Is AI-Generated
Every image, video, carousel, voice track, and caption created in reelmix is generated or modified by artificial intelligence.
- Generated content is labeled as AI-generated in the app, in your library, and in export and publishing screens.
- Characters in reelmix are fictional AI characters.
- Before you publish, we remind you to apply each platform's AI-content label (for example, TikTok's and Instagram's AI labels).
2. Models We Use
The Service routes each task to the model best suited for it. Depending on the task, your content may be created by one or more of the following models.
Images and characters
| Model | Provider | Used for |
|---|---|---|
| Nano Banana Pro (Gemini 3 Pro Image) | Consistent fictional characters, product scenes, readable text on packaging | |
| Seedream 5.0 | ByteDance | Multi-reference scenes and product placement |
| Qwen-Image-Edit 2511 | Alibaba (open model, hosted by reelmix) | Image editing, outfit and scene changes |
| FASHN v1.6 | FASHN AI | Virtual try-on of clothing products |
Video
| Model | Provider | Used for |
|---|---|---|
| Wan 3.0 | Alibaba | Image-to-video animation (primary) |
| Seedance 2.0 | ByteDance | Image-to-video animation (fallback) |
| Veo 3.1 | Image-to-video for close-ups | |
| Kling 2.6 Motion Control | Kuaishou | Character Replacement: transferring motion from a reference video to your fictional AI character |
| SeedVR2 | ByteDance (open model, hosted by reelmix) | Video upscaling |
Voice and lip-sync
| Model | Provider | Used for |
|---|---|---|
| ElevenLabs | ElevenLabs | Voiceovers for fictional characters |
| OmniHuman 1.5 | ByteDance | Talking and lip-synced character video |
| lipsync-2-pro | Sync Labs | Lip-sync correction on generated video |
Text, trends, and analysis
| Model | Provider | Used for |
|---|---|---|
| Claude Haiku 4.5 and Claude Sonnet | Anthropic | Trend classification, scripts, captions, and hooks |
| Whisper (faster-whisper) | OpenAI (open model, hosted by reelmix) | Transcribing reference videos |
Some models are accessed through infrastructure providers such as fal.ai. We update this page when we add, remove, or replace a model.
3. What We Send to Model Providers
- Only the data needed for a task is sent to the provider for that task: your prompt, the relevant uploaded inputs, and reference frames.
- We do not allow providers to train their models on your inputs or outputs where we can control that setting.
- Details are in our Privacy Policy.
4. Safety and Moderation
- Before generation: every prompt is screened in real time by the Waffo content moderation service (Prompt Sift) and our own blocklists.
- After generation: outputs are screened before they are shown to you or published.
- Logs and review: moderation results and human-review decisions are logged.
- No deepfakes: reelmix does not offer face swapping of real people. Character Replacement works only with fictional AI characters created in your account.
- Rules for what you may create are in our Acceptable Use & Content Policy. Report misuse to support@reelmix.ai.
5. Limitations of AI
AI models can produce inaccurate, unexpected, or biased results, and outputs may resemble content generated for other users. Review every output before you publish it. You are responsible for the content you publish.
6. Contact
Questions about our AI models: support@reelmix.ai
YC2123 LLC, doing business as reelmix · 30 N Gould St, Ste R, Sheridan, WY 82801, United States