01
Scope and common rules
This annex applies to the relevant Laintas AI functions together with the Platform Terms, Privacy Policy, Data Use Notice and Billing and Refund Policy. It supplements or overrides general terms only on its stated product matters. Mandatory law and valid individually negotiated agreements take priority. The version expressly provided and validly accepted at purchase or use applies; a web update does not automatically change existing orders.
02
Requests, tools and delivery
Requests may include messages, history, attachments, search and tool results and are processed according to the selected model or route. Citations do not guarantee source accuracy or completeness. Asynchronous media tasks distinguish submission, processing, success and failure; processing is not delivery. Regeneration, continuation and resubmission can create separate calls; duplicate charges and technical retries are assessed against verifiable records.
03
Metering, interruption and charges
Chat generally bills actual model input, output and applicable cached usage; context and internal reasoning may count, so visible characters alone do not reproduce the bill. Media and other capabilities follow the displayed task or unit rate and relevant membership scope. Closing or disconnecting does not necessarily cancel an already submitted supplier task; completed valid work may be billable. Non-delivery, billing errors and refundable failures follow the Billing and Refund Policy rather than being treated as consumption merely because submission occurred.
04
Own keys, source selection and fallback
Activated provider keys follow your source order and capability matching. Calls on your key draw neither Laintas balance nor membership allowance, but the supplier may charge your supplier account. Enabled, authorised fallback sources follow their own rules, including paid balance sources; activation does not make all calls free. Own Compute routes do not silently switch to a paid model because the device is offline or traffic is exhausted. Review source switches and priorities.
05
History, attachments and publication
Chat can remain locally and sync to cloud when Laintas Storage entitlement exists. Attachments and derived content are transmitted to deliver the feature; verify rights and sensitivity before uploading. Conversation deletion, attachment deletion, local clearing and closure have different scopes. Output use and third-party rights follow the Platform Terms; publication follows the AI Content Labelling Rules.
06
Attachments, recognition and external material
Document, image and audiovisual processing can extract text, tables, captions, audio, metadata or excerpts into model context, with possible omissions, misreading or ordering errors. Check source material before relying on contract review, financial calculations, identification or other important processing. Linked pages and files may contain false or malicious instructions; demands to transmit secrets, alter permissions or transact are not your authority. Web and tool content is not Laintas endorsement and does not transfer source rights.
07
Asynchronous tasks, cancellation and resubmission
Accepted media submission means processing has begun; task ID, state and available result together establish delivery. After refresh or timeout, check the original task before assuming a new submission is a free retry. Cancellation stops future work only where actually supported and effective; already submitted upstream work may be irreversible. Distinguish requesting cancellation, actual termination and final billing. Missing downloads, absent content and inconsistent states remain reviewable by task ID despite a success label. Download links, file copies and chat records have different retention; save needed results.
08
Model sources, capabilities and substitution
Models differ in versions, context, tools and media capability. A source label does not establish direct contracting with the brand or official agency. Automatic selection, fallback and retries remain within enabled sources, capability, budget and permissions, respecting explicit confidentiality, funding and data limits. If capability is unavailable, explain failure or alternatives rather than using an unauthorised paid source or broader disclosure. Model updates may alter style and results without permitting arbitrary changes to confirmed prices and promises.