Schedule 2: Other model-specific terms

This Schedule forms part of the TrueOwned Terms of Service. It names the AI models the Service uses besides LTX-2.5 (covered by Schedule 1) and the terms that apply to each. Capitalised terms not defined here have the meaning given in the Terms.

1. Gemma encoder inside LTX-2.5

LTX-2.5 reads your scroll-video instructions through a text encoder derived from Google Gemma 4 12B (file gemma4-12b-with-proj-ltx-2.5-bf16, distributed by Lightricks in the Lightricks/LTX-2.5 repository). No other Gemma model is used: the Service does not use LTX's optional Gemma prompt enhancer or any earlier Gemma generation.

Google licenses Gemma 4 under the Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0), and Google's Gemma Terms of Use direct Gemma 4 users to that licence. The older Gemma Terms of Use and the Gemma Prohibited Use Policy therefore do not apply to this component, and this Schedule adds no separate Gemma use restrictions. The encoder is part of LTX-2.5 as distributed by Lightricks, so the LTX restrictions in Schedule 1 govern the video feature as a whole.

2. Wan 2.2 video fallback

If LTX-2.5 cannot complete a scroll video, the Service may render it with Wan2.2-I2V-A14B (Hugging Face repository Wan-AI/Wan2.2-I2V-A14B-Diffusers, revision 596658fd9ca6b7b71d5057529bbf319ecbc61d74), released by the Wan team of Alibaba under the Apache License 2.0. The fallback is included in the price shown for the video. Receiving a generated video does not mean receiving the model weights. Schedule 1, including its visible-label requirement, applies to every scroll video, and Sections 7 and 8 of the Terms apply as well.

3. Image generation: FLUX.1 [schnell] through fal

AI images are generated by FLUX.1 [schnell] from Black Forest Labs (Apache License 2.0), hosted and run by fal – Features & Labels, Inc. ("fal") on its own infrastructure through the endpoint fal-ai/flux/schnell. For this feature:

1. You must comply with the fal Acceptable Use Policy at https://fal.ai/legal/acceptable-use-policy.
2. You must use the feature and its Output only lawfully and in a way that does not infringe, misappropriate or otherwise violate any rights of fal or any third party.
3. You are responsible for the prompts and other input you submit and for the Output you generate and its later use. Output may not be unique, may be the same as or similar to output generated for others, and may be incorrect; Section 6 of the Terms applies.
4. You must keep the C2PA Content Credentials, IPTC metadata and fal's invisible watermark that the Service applies or preserves (Section 8 of the Terms).
5. The Service accesses fal only through our server. You must not try to obtain or use our fal credentials or to access fal's API through the Service other than through the functions we offer.

4. Website text: Claude 5.x through OpenRouter

Website copy, code and other text are generated by Anthropic Claude 5.5 models (Haiku, Sonnet and Opus), reached through OpenRouter, Inc. and served only by Amazon Bedrock or Google Cloud Vertex AI endpoints that offer zero data retention and do not use requests for training. No separate model-specific use restrictions apply to you beyond Sections 6 to 8 of the Terms. The text carries Anthropic's invisible watermark (Section 8 of the Terms), which you must not deliberately remove or defeat, for example by rewriting text with another tool in order to hide its AI origin.

5. Changes to models and terms

Before a different model or provider restriction applies to you, we will name the affected feature and publish the relevant terms in an updated version of this Schedule, with notice under Section 17 of the Terms. This Schedule does not make every supplier contract a term binding on you; only the terms identified here apply.