Introduction
The Scenario platform provides access to a broad range of AI models for image, 3D, video generation, content editing, custom trainings, and more.
All data is securely hosted and managed on AWS (input/output), and we run compute (generation) directly whenever possible. However, some models are closed-source, and their weights are not publicly available - they can only be accessed through their providers’ APIs. These models are offered through different services, and we’ve compiled an overview of their usage terms below.
For more details on how we handle data and ensure responsible use, please visit trust.scenario.com
License Summary
Image Models
Model | License / Terms | Commercial Use | Data Handling (Input/Output) |
|---|---|---|---|
Ideogram Models | Yes. Users hold rights to generated images and may use them commercially. | Inputs and outputs are not used to train or improve Ideogram’s models. They remain private to the user. | |
Imagen / Nano Banana / Gemini Models | Yes. Outputs may be used commercially without restrictions. | Customer data is not used to train Google’s global models without explicit consent. Inputs/outputs remain under the user’s project control. | |
Seedream 3 | Yes. Generated images can be used commercially, respecting licensing guidelines. | Inputs and outputs are processed securely by the platform. ByteDance retains a broad license, but in enterprise context data is handled privately. | |
Qwen-Image | Yes. Full commercial use, redistribution, and modification are permitted. | Inputs and outputs are processed securely and are not used to train external models. | |
Flux Models | Yes. Outputs, may be used commercially since they are not considered part of the model. | Inputs/outputs are processed securely and not used to train external models. | |
MiniMax Models | Yes. Commercial use is permitted. | Inputs/outputs remain property of the user. Platform processes them securely and does not reuse them for training. | |
Recraft Models | Yes. Users hold full ownership and commercial rights over generated content. | Outputs are private to the user and not reused for training or public display. Inputs are not used to train the Recraft image model. | |
Luma Photon Models | Yes. Users receive outputs without watermarks and full commercial rights. | Outputs are private to the user and not used for training or public distribution. Luma only retains limited rights to host/process data to deliver the service. | |
GPT-Image 1 Models | Yes. Outputs belong to the user, with full commercial rights. | Inputs are not used to train OpenAI models without explicit consent. Data may be temporarily stored (up to 30 days) for abuse monitoring, but not reused for training. | |
Stable Diffusion XL (SDXL) Models | Yes. Broad commercial use is permitted. | Inputs and outputs are processed securely and not reused for training. Outputs belong to the user. | |
Stable Diffusion 1.5 Models | Yes. Commercial use is permitted. | Inputs and outputs are handled securely and not reused. Outputs remain under user ownership. |
Video Models
Model | License / Terms | Commercial Use | Data Handling (Input/Output) |
|---|---|---|---|
Google Veo Models | Yes. Outputs may be used commercially. | Inputs/outputs processed securely; not used for training Google’s global models without consent. | |
Seedance 1 Models | Yes. Generated videos may be used commercially. | Inputs/outputs are processed securely and not reused for training. | |
Wan Models | Yes. Unrestricted commercial use permitted. | inputs/outputs are processed securely and not reused for training. | |
Runway (Gen-4) | Yes. Outputs can be used commercially (ads, films, assets). | Inputs/outputs processed securely to deliver the service; not reused for training in enterprise usage. | |
MiniMax Video Models | Yes. Commercial use is permitted. | Inputs/outputs remain user property, processed securely, not reused for training. | |
Luma Ray 2 Models | Yes. Paid users receive full commercial rights. | Outputs are private and not reused for training. Platform only processes for delivery. | |
Kling Models | Yes. Commercial use supported. | Inputs/outputs processed securely; not reused for training under enterprise context. | |
Pixverse Models | Yes. Outputs may be used commercially. | Outputs remain private to enterprise users; Pixverse does not retain rights to reuse them. | |
FramePack | Yes. Commercial use is permitted. | Inputs/outputs are processed securely and not reused for training. | |
Video Foreground Extractor | Yes. Commercial use is permitted. | Inputs/outputs are processed securely and not reused for training. |
3D Models
Model | License / Terms | Commercial Use | Data Handling (Input/Output) |
|---|---|---|---|
Hunyuan 3D 2.1 | Yes. Commercial use is permitted. | Inputs/outputs processed securely, not reused for training. | |
Hunyuan 3D 2.0 | Yes. Broadly usable under Apache 2.0-like conditions. | Inputs/outputs processed securely, not reused for training. | |
PartCrafter | Yes. Open-source under MIT License; unrestricted commercial use. | Inputs/outputs are processed securely and not reused for training. | |
Tripo 2.5 Models | Yes. Generated 3D models may be used commercially in products, games, or videos. | Inputs processed securely, outputs belong to user, not reused for training. | |
Rodin Hyper3D Models | Yes. Outputs can be used commercially without restrictions. | Inputs/outputs processed securely; not reused for training. | |
Direct3D-S2 | Yes. Commercial use allowed. | Inputs/outputs are processed securely and not reused for training. | |
Trellis (Microsoft) | Yes. Commercial use allowed. | Inputs/outputs are processed securely and not reused for training. |
Audio
Model | License / Terms | Commercial Use | Data Handling (Input/Output) |
|---|---|---|---|
ElevenLabs Multilingual v2 | Yes. Commercial use permitted under ElevenLabs terms. | Inputs securely processed by ElevenLabs; outputs returned to Scenario. Not reused for training. | |
ElevenLabs Sound Effects | Yes. Commercial use permitted under ElevenLabs terms. | Inputs securely processed by ElevenLabs; outputs returned to Scenario. Not reused for training. | |
ElevenLabs Turbo v2.5 / Flash v2.5 | Yes. Commercial use permitted. | Inputs securely processed by ElevenLabs; outputs returned to Scenario. Not reused for training. | |
MiniMax Music-01 | Yes. Commercial use permitted under MiniMax terms. | Inputs securely processed by MiniMax; outputs returned to Scenario. Not reused for training. | |
MiniMax Speech-02 HD | Yes. Commercial use permitted under MiniMax terms. | Inputs securely processed by MiniMax; outputs returned to Scenario. Not reused for training. | |
MiniMax Speech-02 Turbo | Yes. Commercial use permitted. | Inputs securely processed by MiniMax; outputs returned to Scenario. Not reused for training. | |
MiniMax Voice Cloning | Yes. Commercial use permitted; cloned voices expire after 168h if unused. | Inputs securely processed by MiniMax; temporary voice IDs deleted automatically after expiration. Outputs returned to Scenario. | |
ByteDance OmniHuman-1 | Restricted. Not publicly licensed for commercial use; currently research/internal only. | Inputs securely processed by ByteDance; outputs returned to Scenario. Not reused for training. | |
Creatify LipSync | Yes. Commercial use permitted under Creatify terms. | Inputs securely processed by Creatify; outputs returned to Scenario. Not reused for training. | |
Kling Lip Sync (Kuaishou) | In-app / platform terms | Yes. Commercial use permitted under Kuaishou terms. | Inputs securely processed by Kuaishou; outputs returned to Scenario. Not reused for training. |
PixVerse LipSync | Yes. Commercial use permitted under PixVerse license. | Inputs securely processed by PixVerse; outputs returned to Scenario. Not reused for training. | |
Sync Labs v2 (LipSync-2) | Yes. Commercial use permitted under Sync Labs terms. | Inputs securely processed by Sync Labs; outputs returned to Scenario. Not reused for training. | |
Google Lyria 2 | Restricted. Non-commercial or research use emphasized under Google’s terms. | Inputs securely processed by Google; outputs returned to Scenario. Not reused for training. | |
Meta MusicGen (AudioCraft) | Non-commercial license (community release) | No. Non-commercial only. | Inputs/outputs processed securely by Scenario, not reused for training. |
Resemble AI Chatterbox | Yes. MIT License allows commercial use; subject to Resemble AI terms. | Inputs/outputs processed securely by Scenario, not reused for training. |
Editing
Editing Model | License / Terms | Commercial Use | Input / Output Handling |
|---|---|---|---|
Luma Reframe (video reframing/outpainting) | Yes. Users retain commercial usage rights. | Inputs and outputs are processed securely and remain private. | |
Runway Aleph (in-context video editing with text prompts) | Yes. Users retain full ownership and commercial rights. | Inputs and outputs are processed securely to provide editing services; ownership remains with user throughout. | |
Topaz Video Upscaler (Video AI Pro) | Yes. License terms allow commercial use. | Inputs and outputs are processed securely. | |
Gemini 2.5 / Nano Banana — Edit with Prompts | Yes. Commercial use permitted under Google API terms. | Inputs and edited outputs are processed securely. Google’s policy states user content is not used for training without consent. | |
Flux Kontext | Yes. Commercial use allowed under API terms. | Inputs and prompts are processed securely for iterative editing via text + images. Ownership stays with user. | |
GPT-Image — Edit with Prompts | Yes. Full commercial usage rights. | Inputs are not used to train OpenAI models by default; retained only up to 30 days for abuse prevention. | |
Runway Gen-4 Edit with Prompts | Yes. You retain ownership and can use output commercially. | Uploaded images and instructions are processed securely to transform the image; you retain ownership of both inputs and outputs. |
Others
Model | License / Terms | Commercial Use | Data Handling (Input/Output) |
|---|---|---|---|
Flux LoRAs | Yes. Commercial license covers use of Flux and LoRA fine-tunes for business purposes. | Inputs and outputs are processed securely and not reused for training. Users retain ownership of generated results. | |
SDXL LoRAs | Yes. Covered by commercial license; SDXL LoRA models can be used commercially. | Inputs and outputs are private, not reused for training. Outputs belong to the user. | |
SD 1.5 Models & LoRAs | Yes. Commercial use permitted under RAIL-M. LoRAs inherit the same licensing. | Inputs/outputs are processed securely, not reused for training. User retains ownership of results. |
Conclusion
The models listed above are cleared for commercial use. They can be integrated into image, video, and 3D generation workflows with full compliance under their respective licenses. This gives both individual creators and enterprise teams the freedom to build and scale projects without restrictions on deployment.
Was this helpful?