Question 1

OmniHuman 1.5 via Hedra: Commercial use?

Accepted Answer

Restricted. Source: ByteDance's terms, read 2026-10-21. "Service Specific Terms s4.2.3 (Video Generation, 2026-09-30): you may not 'resell or re-integrate the APIs into UGC platforms, including AI content creation tools. If you operate an aggregation platform, end users may only generate video content for their own use'; nor 'resell the APIs without our written authorization'. The platform is liable for its end users (s4.2.2); unauthorised resale costs 20% of the minimum commitment (s4.2.4). Client / agency deliverables through a channel are UNCLEAR (asked). Model Services s2(1): not available in the United States - whether channels serve US end users under a separate arrangement is unconfirmed. Ownership: you own the Output as between you and BytePlus (General Terms for AI Services s2.1)."

Question 2

OmniHuman 1.5 via Hedra: Where outputs may be used?

Accepted Answer

Restricted. Source: ByteDance's terms, read 2026-10-21. "Service Specific Terms s4.2.3 (Video Generation, 2026-09-30): you may not 'resell or re-integrate the APIs into UGC platforms, including AI content creation tools. If you operate an aggregation platform, end users may only generate video content for their own use'; nor 'resell the APIs without our written authorization'. The platform is liable for its end users (s4.2.2); unauthorised resale costs 20% of the minimum commitment (s4.2.4). Client / agency deliverables through a channel are UNCLEAR (asked). Model Services s2(1): not available in the United States - whether channels serve US end users under a separate arrangement is unconfirmed. Ownership: you own the Output as between you and BytePlus (General Terms for AI Services s2.1)."

Question 3

OmniHuman 1.5 via Hedra: Trains on your inputs?

Accepted Answer

No. Source: ByteDance's terms, read 2026-10-21. "Service Specific Terms, ModelArk s4 (2026-09-30): 'Unless you separately opt in, BytePlus will not use your data to train models.' General Terms for AI Services s3.3 (2026-08-28): 'Except for feedback that you voluntarily provide for service improvement, or where you otherwise separately consent or agree in writing, BytePlus will not use your Customer Data to train the foundation models underlying the AI Services.' Video terms: the Video Generation Model Services will not use the prompts and other input you submit, or the output generated, to train the base models unless you separately provide consent."

Question 4

OmniHuman 1.5 via Hedra: Trains on your prompts?

Accepted Answer

No. Source: ByteDance's terms, read 2026-10-21. "Prompts are the defined Input. Service Specific Terms, ModelArk s4 (2026-09-30): 'Unless you separately opt in, BytePlus will not use your data to train models.' General Terms for AI Services s3.3 (2026-08-28): 'Except for feedback that you voluntarily provide for service improvement, or where you otherwise separately consent or agree in writing, BytePlus will not use your Customer Data to train the foundation models underlying the AI Services.' Video terms: the Video Generation Model Services will not use the prompts and other input you submit, or the output generated, to train the base models unless you separately provide consent."

Question 5

OmniHuman 1.5 via Hedra: Trains on your outputs?

Accepted Answer

No. Source: ByteDance's terms, read 2026-10-21. "Output named explicitly in the Video terms. Service Specific Terms, ModelArk s4 (2026-09-30): 'Unless you separately opt in, BytePlus will not use your data to train models.' General Terms for AI Services s3.3 (2026-08-28): 'Except for feedback that you voluntarily provide for service improvement, or where you otherwise separately consent or agree in writing, BytePlus will not use your Customer Data to train the foundation models underlying the AI Services.' Video terms: the Video Generation Model Services will not use the prompts and other input you submit, or the output generated, to train the base models unless you separately provide consent."

Question 6

OmniHuman 1.5 via Hedra: Copyright indemnity?

Accepted Answer

Limited. Source: ByteDance's terms, read 2026-10-21. "Service Specific Terms, Model Services s3.1 (2026-09-30): 'BytePlus shall defend you against a third party claim alleging that your permitted use of any Output generated by a BytePlus Model Services, which is NOT using third party models, BytePlus Modified Models or Third-party Modified Model, infringes the intellectual property rights of that third party, and shall indemnify you for all amounts awarded against you in final judgment or settlement'. s3.2 excludes: knew or should have known; bypassed safety features; modified or combined Output; trademark use; ignored instructions; use after notice; fraud, negligence or breach; Early Access / Preview models. No cap stated. Tension: s3.3(e) makes you indemnify BytePlus against 'any allegation that your... use of Outputs infringes'. Whether the indemnity extends to end users of channel partners is unconfirmed (asked)."

Question 7

OmniHuman 1.5 via Hedra: Marking applied by the model?

Accepted Answer

Yes. Source: ByteDance's terms, read 2026-10-21. "C2PA Content Credentials Guide (2026-07-28): the ModelArk API 'embeds C2PA Content Credentials by default... No additional configuration is required' for ByteDance-Seedream-4.0, Seedream-4.5, Dola-Seedream-5.0-pro / -lite, Dreamina-Seedance-2.0 / -2.0-mini / -2.0-fast (Seedance 2.5 announced, not listed; Seedream 3.0 and Seedance 1.x not covered). Acceptable Use Policy: users must not 'delete, remove, obscure, alter or conceal any credentials, watermarks, identifiers, metadata... that indicate that the content may be generated by a BytePlus GenAI Service'. General Terms for AI Services s1.4: comply with AI labelling / disclosure law when publishing. Visible watermark default: unconfirmed."

Question 8

OmniHuman 1.5 via Hedra: Disclosure duty on you?

Accepted Answer

Required. Source: ByteDance's terms, read 2026-10-21. "C2PA Content Credentials Guide (2026-07-28): the ModelArk API 'embeds C2PA Content Credentials by default... No additional configuration is required' for ByteDance-Seedream-4.0, Seedream-4.5, Dola-Seedream-5.0-pro / -lite, Dreamina-Seedance-2.0 / -2.0-mini / -2.0-fast (Seedance 2.5 announced, not listed; Seedream 3.0 and Seedance 1.x not covered). Acceptable Use Policy: users must not 'delete, remove, obscure, alter or conceal any credentials, watermarks, identifiers, metadata... that indicate that the content may be generated by a BytePlus GenAI Service'. General Terms for AI Services s1.4: comply with AI labelling / disclosure law when publishing. Visible watermark default: unconfirmed."

Can I use OmniHuman 1.5 via Hedra commercially?

Sources

OmniHuman 1.5 on other services

Other models on Hedra

Recent changes to these terms

Tell me when these terms change

Keep this record for your own work

Question	Answer	According to	Read on
Commercial use	Restricted Own use only	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"Service Specific Terms s4.2.3 (Video Generation, 2026-09-30): you may not 'resell or re-integrate the APIs into UGC platforms, including AI content creation tools. If you operate an aggregation platform, end users may only generate video content for their own use'; nor 'resell the APIs without our written authorization'. The platform is liable for its end users (s4.2.2); unauthorised resale costs 20% of the minimum commitment (s4.2.4). Client / agency deliverables through a channel are UNCLEAR (asked). Model Services s2(1): not available in the United States - whether channels serve US end users under a separate arrangement is unconfirmed. Ownership: you own the Output as between you and BytePlus (General Terms for AI Services s2.1)."
Where outputs may be used	Restricted not offered in US	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"Service Specific Terms s4.2.3 (Video Generation, 2026-09-30): you may not 'resell or re-integrate the APIs into UGC platforms, including AI content creation tools. If you operate an aggregation platform, end users may only generate video content for their own use'; nor 'resell the APIs without our written authorization'. The platform is liable for its end users (s4.2.2); unauthorised resale costs 20% of the minimum commitment (s4.2.4). Client / agency deliverables through a channel are UNCLEAR (asked). Model Services s2(1): not available in the United States - whether channels serve US end users under a separate arrangement is unconfirmed. Ownership: you own the Output as between you and BytePlus (General Terms for AI Services s2.1)."
Trains on your inputs	No	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"Service Specific Terms, ModelArk s4 (2026-09-30): 'Unless you separately opt in, BytePlus will not use your data to train models.' General Terms for AI Services s3.3 (2026-08-28): 'Except for feedback that you voluntarily provide for service improvement, or where you otherwise separately consent or agree in writing, BytePlus will not use your Customer Data to train the foundation models underlying the AI Services.' Video terms: the Video Generation Model Services will not use the prompts and other input you submit, or the output generated, to train the base models unless you separately provide consent."
Trains on your prompts	No	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"Prompts are the defined Input. Service Specific Terms, ModelArk s4 (2026-09-30): 'Unless you separately opt in, BytePlus will not use your data to train models.' General Terms for AI Services s3.3 (2026-08-28): 'Except for feedback that you voluntarily provide for service improvement, or where you otherwise separately consent or agree in writing, BytePlus will not use your Customer Data to train the foundation models underlying the AI Services.' Video terms: the Video Generation Model Services will not use the prompts and other input you submit, or the output generated, to train the base models unless you separately provide consent."
Trains on your outputs	No	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"Output named explicitly in the Video terms. Service Specific Terms, ModelArk s4 (2026-09-30): 'Unless you separately opt in, BytePlus will not use your data to train models.' General Terms for AI Services s3.3 (2026-08-28): 'Except for feedback that you voluntarily provide for service improvement, or where you otherwise separately consent or agree in writing, BytePlus will not use your Customer Data to train the foundation models underlying the AI Services.' Video terms: the Video Generation Model Services will not use the prompts and other input you submit, or the output generated, to train the base models unless you separately provide consent."
Output ownership	not on record	-	-
Copyright indemnity	Limited indemnity runs to the API customer	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"Service Specific Terms, Model Services s3.1 (2026-09-30): 'BytePlus shall defend you against a third party claim alleging that your permitted use of any Output generated by a BytePlus Model Services, which is NOT using third party models, BytePlus Modified Models or Third-party Modified Model, infringes the intellectual property rights of that third party, and shall indemnify you for all amounts awarded against you in final judgment or settlement'. s3.2 excludes: knew or should have known; bypassed safety features; modified or combined Output; trademark use; ignored instructions; use after notice; fraud, negligence or breach; Early Access / Preview models. No cap stated. Tension: s3.3(e) makes you indemnify BytePlus against 'any allegation that your... use of Outputs infringes'. Whether the indemnity extends to end users of channel partners is unconfirmed (asked)."
Marking applied by the model	Yes documentation, not contract	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"C2PA Content Credentials Guide (2026-07-28): the ModelArk API 'embeds C2PA Content Credentials by default... No additional configuration is required' for ByteDance-Seedream-4.0, Seedream-4.5, Dola-Seedream-5.0-pro / -lite, Dreamina-Seedance-2.0 / -2.0-mini / -2.0-fast (Seedance 2.5 announced, not listed; Seedream 3.0 and Seedance 1.x not covered). Acceptable Use Policy: users must not 'delete, remove, obscure, alter or conceal any credentials, watermarks, identifiers, metadata... that indicate that the content may be generated by a BytePlus GenAI Service'. General Terms for AI Services s1.4: comply with AI labelling / disclosure law when publishing. Visible watermark default: unconfirmed."
Disclosure duty on you	Required	ByteDance terms (model provider) docs.byteplus.com	2026-10-21
	"C2PA Content Credentials Guide (2026-07-28): the ModelArk API 'embeds C2PA Content Credentials by default... No additional configuration is required' for ByteDance-Seedream-4.0, Seedream-4.5, Dola-Seedream-5.0-pro / -lite, Dreamina-Seedance-2.0 / -2.0-mini / -2.0-fast (Seedance 2.5 announced, not listed; Seedream 3.0 and Seedance 1.x not covered). Acceptable Use Policy: users must not 'delete, remove, obscure, alter or conceal any credentials, watermarks, identifiers, metadata... that indicate that the content may be generated by a BytePlus GenAI Service'. General Terms for AI Services s1.4: comply with AI labelling / disclosure law when publishing. Visible watermark default: unconfirmed."
Training data disclosed	not on record	-	-