Effective Date: October 14, 2026
Last Updated: October 14, 2026
Service Provider: Max Chao (趙士翔)
Location: Taipei, Taiwan
Contact Email: zawshishawn@gmail.com
Welcome to Gestura (the "App"). These Terms of Service (the "Terms") govern the rights and obligations between you and the individual developer Max Chao (趙士翔) ("we," "us," or "our") when you download, install, access, browse, or use the App.
By downloading, installing, opening, or using the App, you acknowledge that you have read, understood, and agreed to be bound by these Terms. If you do not agree, do not download, install, or use the App. Where applicable law requires separate express consent, we will follow that requirement.
Gestura is a mobile application that provides text and image processing, AI-generated text, on-device image generation, on-device video generation with audio from supported text, image, video, and audio inputs, on-device music generation and music transcription, model browsing and downloads, offline AI execution, and connections to user-configured servers or API endpoints. Available features may vary by App version, device, operating system, region, model, hardware capabilities, or third-party service.
The App may include the following modes or features:
Offline mode. Text, images, video, audio, model inference, and other supported AI processing may be performed locally on your device. The App does not transmit local inputs, prompts, or outputs to us or an external server merely to perform that local processing.
On-device generation and music transcription. You may use compatible models stored on your device to generate images, videos with audio, or music, or to perform music transcription. The MiniMax H3-related video feature accepts supported text, image, video, and audio inputs and performs generation locally on your mobile device; it is not a MiniMax or Hugging Face hosted-inference service. Network activity for model downloads and other separately selected features is distinct from local generation and transcription.
Connected mode or custom-server mode. You may configure the App to connect to an Ollama server, another custom API endpoint, or a third-party service. In this mode, text, images, prompts, context, or other request data you submit may be transmitted to the server or API you select to generate text, images, or other AI outputs.
Model lists and downloads. The App may use the Hugging Face API or other public sources to retrieve model lists, names, descriptions, authorship, license information, file information, or download links. For the H3-related feature, you download the developer-provided quantized models and related components from Hugging Face through the App, subject to Section 11.
Reports and feedback. You may submit a text report together with App version, device model, Android version, and system language. These data are sent only when you press the submit control. In-app reports do not accept image, video, audio, or thumbnail attachments, and the App does not automatically upload generation or transcription records or diagnostic data.
Voluntary support. The App may offer voluntary support payments through in-app purchases. A support payment does not unlock additional features, content, privileges, models, badges, priority support, or other digital goods.
Please also review the Gestura Privacy Policy for information about the App's data practices.
You must satisfy the age and legal-capacity requirements under the laws of your location to use the App and agree to these Terms.
The App is not directed to children under 13. If you are under 13, do not use the App. If the law in your location imposes a higher minimum age or requires parental or guardian consent, that requirement applies. In the European Union and European Economic Area, the age at which a child may independently consent to personal-data processing for information society services may vary from 13 to 16 under Member State law.
If you do not have full legal capacity to enter into these Terms, you may use the App only with the consent and supervision of a parent, legal guardian, or other person legally authorized to act for you, and that person must review and agree to these Terms on your behalf.
Subject to your compliance with these Terms and applicable law, we grant you a limited, non-exclusive, non-transferable, non-sublicensable, revocable license to download, install, and use the App solely on devices that you personally own or lawfully control.
Unless expressly permitted by applicable law, an open-source license, or our prior written consent, you may not:
copy, modify, translate, adapt, distribute, sell, rent, transfer, or sublicense the App;
reverse engineer, decompile, disassemble, or attempt to discover non-public source code, algorithms, or trade secrets of the App;
remove, obscure, or alter any copyright, trademark, open-source license, attribution, or other rights notice in the App;
use the App to create a substantially competing product or service, except to the extent expressly permitted by applicable law or an open-source license; or
use the App in a manner that may damage, interfere with, disable, or overburden the App, a third-party API, a server, a model source, or a related system.
Third-party and open-source components included in the App remain governed by their respective licenses. These Terms do not replace, restrict, or reduce any rights or obligations under those licenses.
You are solely responsible for content that you input, submit, process, transmit, download, generate, store, export, publish, or use through the App, including text, prompts, reference images, source video and audio, personal likenesses and voices, recordings, music, lyrics, scores, performances, AI-generated text, generated images, videos and their audio tracks, generated music, transcription outputs, model files, reports, and feedback.
You represent and warrant that you have all ownership rights, licenses, consents, permissions, or other lawful bases required for that content and that your use will not violate law, these Terms, third-party rights, model licenses, platform policies, or third-party service terms.
You must take reasonable measures to protect your device, models, API keys, server addresses, inputs, outputs, and other data. You are responsible for deciding whether sensitive, confidential, personal, or protected information is appropriate to enter into the App or transmit to a third-party service.
The App may process text, image, video, and audio inputs supported by the selected feature and produce AI-generated text, images, videos with audio, music, or transcription outputs based on the model, settings, and computing environment. AI outputs are probabilistic and do not constitute a statement, endorsement, or guarantee by us, a model provider, or any third party.
You understand and agree that:
AI-generated text, images, videos, audio tracks, music, or transcription outputs may be inaccurate, incomplete, outdated, biased, offensive, unsafe, inconsistent with your prompt, or unsuitable for a particular purpose;
outputs may contain fabricated information or speech, distorted persons or objects, incorrect text, visual or audio defects, temporal inconsistencies, or audio-video synchronization errors, and may resemble existing works, voices, recordings, music, performances, persons, trademarks, characters, styles, or other content;
the same or similar prompts may produce different results, and different users may receive identical or similar outputs; outputs are not guaranteed to be unique;
AI outputs must not be treated as facts, reliable records of real events, professional opinions, legal advice, medical advice, financial advice, investment advice, or other advice requiring professional judgment;
before using, publishing, sharing, selling, licensing, distributing, or relying on an output, you must independently review, evaluate, verify, and obtain any necessary rights or consents; and
you assume the risks arising from your use, reliance, modification, publication, distribution, or commercial exploitation of AI outputs.
We do not warrant the accuracy, completeness, originality, legality, copyrightability, suitability, safety, reproducibility, commercial usability, or non-infringement of any AI output. Music transcription may contain pitch, rhythm, or note-recognition errors; you must check and correct the results before use.
On-device image, video, and music generation and music transcription use models, supported inputs, prompts, settings, and computing resources on your device. The H3-related video feature processes text, images, video, and audio locally to generate videos with audio. Local generation and transcription do not upload those inputs or outputs to us, MiniMax, Hugging Face, or another external server. A model download is a separate network operation, not an upload of your generation or transcription content.
You understand and agree that:
model downloads, App updates, app-store functions, license-information retrieval, custom endpoints, reports, or sharing may still require a network connection;
on-device image, video, and music generation and music transcription may make intensive use of the CPU, GPU, NPU, memory, storage, battery, and other device resources and may cause heat, battery drain, thermal throttling, termination of background processes, or other performance effects;
generation and transcription speed, video and music duration, resolution, frame rate, image and audio quality, audio-video synchronization, transcription quality, stability, and available models depend on device capabilities, operating system, model format, model size, available memory, storage, and other technical conditions;
prompts, inputs, settings, previews, caches, temporary files, and outputs may be stored locally according to the App's design and your actions. Generated videos are saved within the App by default, not automatically to your photo library. When you press the download or export control, a video is saved to your photo library. Exported files may remain after you uninstall the App;
generated files and transcription outputs may contain timestamps, dimensions, duration, format, or other technical metadata. Before sharing, you must consider privacy and any obligation to preserve attribution, AI-generation disclosures, watermarks, or provenance information; and
content filters, safety checks, notices, or warnings cannot guarantee that all prohibited content will be blocked. Their presence or failure to block content does not mean that an output is lawful, appropriate, reviewed, or licensed. You must not tamper with files, crack protections, or otherwise bypass safeguards to generate prohibited content.
You are responsible for managing, backing up, and deleting locally generated content and transcription outputs and for taking appropriate data-protection measures before your device is lost, shared, serviced, sold, or disposed of.
You retain the rights you already hold in prompts, source images, video, audio, and other inputs. We do not claim ownership of images, videos, their audio tracks, music, or transcription outputs produced entirely on your device and not voluntarily submitted to us. To the extent we might have any rights in those outputs under applicable law, we do not assert those rights. This does not mean that you necessarily obtain copyright, other exclusive rights, or unrestricted usage rights in an output.
Your ability to use generated images, videos, audio tracks, music, and transcription outputs may be affected by:
the rights status of the prompt, source image, video, audio, recording, music, lyrics, score, performance, reference material, or other input;
the license terms and acceptable-use policies of the model, model weights, add-ons, data, or related tools;
the terms of a custom server, third-party API, app store, or other platform;
applicable copyright, trademark, publicity, personality, privacy, personal-data, trade-secret, consumer-protection, advertising, election, export-control, or other laws; and
whether the output is identical or similar to a third-party work, character, brand, person, product, place, or other protected subject matter.
You are responsible for any required rights clearance, similarity review, license verification, and legal review, including permissions for voices, recordings, music, lyrics, scores, and performances. We do not warrant that generated images, videos, audio tracks, music, or transcription outputs can be registered, protected by copyright, commercially used, sublicensed, redistributed, used as a trademark, or used without a third-party claim. Transcription does not grant you rights in the original music, recording, or score or replace any license or other lawful basis required to use or distribute the transcription output.
If a model license, law, platform policy, or transaction counterparty requires disclosure that content is AI-generated, retention of source information, model attribution, or preservation of content credentials, watermarks, or provenance information, you must comply and must not unlawfully or deceptively remove, falsify, or conceal such information. The H3-specific conditions in Section 11 also apply to H3-related content.
If you input, generate, edit, or publish images, videos, or audio containing an identifiable person, face, body, voice, speech, name, identity characteristic, intimate context, or other personality-related information, you must have the rights, consent, or other lawful basis required by applicable law. Supplying a person's photo, video, or voice recording does not by itself establish their consent to synthetic depiction or imitation.
You may not use the App to:
create, modify, possess, or distribute child sexual exploitation or child sexual abuse material, or content that sexualizes a minor;
create or distribute non-consensual intimate imagery, sexual imagery, or synthetic nudity, or use such content to threaten, extort, humiliate, or harass another person;
impersonate another person, falsify identity, commit fraud, create an unauthorized endorsement, or materially mislead others about the source, authenticity, conduct, statements, events, or relationships depicted in a manner likely to cause substantial harm;
fabricate evidence, forge official documents, or deceptively influence judicial, administrative, employment, education, financial, insurance, election, or other significant processes;
violate another person's publicity, personality, privacy, reputation, personal-data, or other legal rights; or
target a person or group through unlawful hatred, discrimination, violence, harassment, stalking, or abuse.
Where required by applicable law, model license, platform policy, professional standards, or the need to avoid reasonable misunderstanding, you must clearly disclose that content is AI-generated or AI-modified. Images, videos, and audio generated by the App must not be treated as reliable proof of a real person, speech, performance, event, document, or scene.
In offline mode, the App processes supported text, images, video, audio, model inference, image generation, video generation with audio, music generation, music transcription, or other AI computation locally on your device. You are responsible for ensuring that the device has sufficient performance, memory, storage, battery, security settings, and a compatible environment.
We do not guarantee that offline mode or on-device image, video, or music generation or music transcription will function on every device, Android version, language, model format, model size, hardware configuration, or system environment. Restrictions imposed by a device manufacturer, operating system, or another app may also affect functionality.
You are responsible for backing up important data. If device failure, model error, system update, insufficient storage, user error, file corruption, permission changes, or another cause results in lost data, images, videos or audio, damaged models, or an unusable App, we are responsible only to the extent liability cannot lawfully be excluded.
When you use connected mode and configure the App to connect to an Ollama server, custom API endpoint, or other third-party service, you understand and agree that:
text, images, prompts, context, settings, source images, or other request data you submit may be transmitted to and processed or retained by the server or API you select;
we do not control the server, API, model, third-party service, its security measures, or its data practices;
you are responsible for verifying endpoint addresses, credentials, API keys, encryption, security, privacy policies, terms, data retention, content policies, charges, rate limits, and availability;
you may not use an account, credential, API key, computing resource, or server without authorization, and you must protect keys and connection information from disclosure;
you assume the risks of data leakage, unauthorized access, service interruption, charges, erroneous outputs, content restrictions, or other harm arising from your connection, configuration, or use; and
we are not responsible for any third-party server, API, model, service, content, payment, or conduct.
The H3-related on-device video feature described in Section 6 does not use this connected mode to perform generation. Merely configuring a custom endpoint does not change that feature into cloud video generation.
Ollama, Hugging Face, MiniMax, MiniMax H3, llama.cpp, and other third-party names, trademarks, service names, or project names belong to their respective owners. Unless expressly stated otherwise, the App is not an agent, partner, employee, authorized reseller, endorsee, or affiliate of those third parties.
The App may use the Hugging Face API or other public sources to retrieve lists, descriptions, authorship, license labels, file information, or download links for text models, image-generation models, video-generation models with audio, music-generation models, music-transcription models, and related components. Components may include model weights, quantized files, delta files, text or multimodal encoders, video or audio decoders, VAEs, LoRAs, adapters, tokenizers, safety components, or other files.
You understand and agree that:
models, model weights, data, code, model cards, documentation, and other materials may be provided by different authors, organizations, or rights holders;
different models and components may be subject to different licenses, use restrictions, acceptable-use policies, research-only or non-commercial restrictions, attribution requirements, distribution conditions, or output-related conditions;
whether a model may be commercially used, modified, redistributed, embedded in a product, offered as a service, or used for a particular type of content is governed by the model page, model card, license document, and rights-holder terms;
displaying a model or providing a download entry does not mean that we own, review, recommend, warrant the safety or legality of, or grant you any particular right to use that model;
before downloading, running, modifying, redistributing, or commercially using a model, you must read and comply with all applicable terms. If the license is unclear or incompatible with your intended use, you must not use the model or component;
you may not remove, obscure, or alter copyright, license, source, restriction, or other rights notices in a model, model card, file, license document, or third-party content; and
third-party models may contain errors, malicious code, unsafe content, bias, technical defects, or incompatible files and may consume substantial bandwidth, storage, or computing resources. You are responsible for evaluating source and risk.
You are responsible for disputes, claims, harm, costs, or liabilities attributable to your download, use, modification, redistribution, service offering, or commercial exploitation of a third-party model or component, to the extent permitted by law. This does not exclude or transfer to you our own obligations or liabilities as the App provider or distributor of modified model files.
For the H3-related video feature, the App uses developer-provided quantized models and related components downloaded from Hugging Face through the App.
The general modification notices for the Gestura H3 model packages state that all model payloads listed in the applicable files.json manifest were modified by Gestura through conversion, quantization, or delta packaging. These payloads are not unmodified upstream releases. This declaration concerns the listed model payloads; other files and components remain subject to their own applicable notices.
The original model is MiniMax H3. Applicable source acknowledgments and rights notices for the model packages and related components are provided in the App's shared licensing section. No upstream endorsement of Gestura or its modifications is implied.
For separately licensed third-party contributions identified as Apache-2.0 in the App's shared licensing section, that license applies only to the rights and material actually licensed by the relevant licensor. It does not relicense MiniMax H3 or remove the H3 agreement's conditions for H3 works and derivatives. Applicable Apache-2.0 license-copy, modification-notice, and attribution requirements must also be observed.
The permissions and restrictions of every applicable upstream and component license remain in force. Quantization, conversion, packaging, or free availability does not remove them. A repository name, download entry, or summary is not a substitute for the applicable license files and notices.
The App includes the full MiniMax H3 Community License Agreement and Exhibit A (Acceptable Use Policy). During the standard model-download flow, the App displays those documents and requires you to check the agreement box and accept before continuing with the relevant model download.
Before downloading or using the H3-related feature, you must read and agree to comply with the MiniMax H3 Community License Agreement, including Section V and Exhibit A (Acceptable Use Policy). Those use restrictions are incorporated into these Terms and bind you when you access or use that feature. The summary below is not exhaustive. The full agreement governs the licensed H3 works, and these Terms grant no additional model or output rights.
https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE
Under the license dated August 2, 2026, the excluded territories are the European Union, United Kingdom, Republic of Korea, and United States. The restrictions cover use, reproduction, modification, distribution, and display of the H3 works and their outputs in those territories. Public dissemination of H3-generated content requires clear, prominent AI-generation disclosure. H3 works or outputs must not improve other AI models, except H3 or its derivatives.
Before the App permits the relevant model download, it asks whether you are in a restricted territory. If you answer that you are, the App does not make that download available. The answer is processed only on your mobile device and is not sent to us or third parties as part of that declaration.
You must answer accurately and must not bypass the restriction through a false declaration or file tampering. Passing this download check or previously downloading a model does not establish that later model use or distribution of outputs is authorized. These H3 conditions do not replace the separate licenses governing other models.
The App's shared licensing section includes a model-source document containing the H3 NOTICE statement and a general declaration that assets converted by Gestura are modified versions. You can review that document in the shared licensing section; the current pre-download dialog does not load it. The license-copy, NOTICE-text-file, and modification-notice requirements below remain applicable regardless of the display location.
For H3 works that we distribute or make available, we remain responsible for providing recipients with the full H3 agreement, ensuring that modified files carry prominent modification notices, and accompanying non-hosted distributions with the required NOTICE text file, in accordance with Section III of that agreement. The required notice reads:
MiniMax H3 is licensed under the MiniMax H3 Community License Agreement, Copyright © 2026 MiniMax. All Rights Reserved.
Applicable licenses and upstream notices must also be preserved for separately licensed components. A model-card label, a source URL, or this summary does not replace a required license copy or NOTICE file. If you further distribute H3 works, the full H3 agreement governs the distribution requirements and any applicable exception for recipients of an integrated end-user product; its Section V and Exhibit A continue to apply.
These Terms do not replace our responsibilities under Section V.5 of the H3 agreement to implement, maintain, test, and periodically review reasonable and proportionate safeguards, maintain an accessible reporting mechanism, and investigate and reasonably address violations within our control. Local operation and user responsibilities do not remove those provider obligations.
The App may use third-party software, open-source components, SDKs, APIs, model runtimes, model sources, app stores, or platform services. Those components and services remain governed by their respective licenses, terms, and platform policies, and these Terms do not replace, restrict, or modify them.
For lists of open-source and third-party components, copyright notices, license texts, NOTICE files, and disclaimers, review the "Open-Source Licenses" page in the App or on the applicable page. If a third-party or open-source license differs from these Terms, that license governs the applicable component, SDK, API, model, or third-party service itself.
The App license in Section 3 applies to Gestura software, not as a replacement license for the H3 works or separately licensed components. Nothing in these Terms imposes additional or different terms on H3 modifications contrary to the H3 agreement.
The App offers a text-based "Report," "Feedback," or similar feature. We receive the report and accompanying device and system information only when you actively press the submit control. The App does not automatically upload generation or transcription records or diagnostic data, and generating an image, video, or music or performing transcription does not send the media, transcription output, or a thumbnail to us.
The in-app report submission is limited to:
text you choose to submit, such as feedback, a problem description, or AI text and prompts that you deliberately include; and
App version, device model, Android version, and system language, sent with the text only when you submit the report.
The in-app report feature does not accept image, video, audio, or thumbnail attachments. Model details, settings, or error information are included only to the extent you choose to describe them in the report text; this feature does not automatically upload other diagnostic files.
You grant us a worldwide, non-exclusive, royalty-free, limited license to store, use, reproduce, analyze, and process the text that you voluntarily submit solely to diagnose and fix errors, improve stability and performance and the reliability of AI features, provide language support, respond to your request, address reported misuse, and comply with legal obligations. This license does not extend to media or other content kept only on your device.
Do not include unnecessary personal data, intimate details, confidential information, third-party secrets, or content you are not authorized to provide. Unless required by law or separately agreed by you, we do not use submitted data for advertising, marketing, cross-service tracking, advertising profiling, or sale to third parties. Retention, deletion requests, and other data rights are described in the Gestura Privacy Policy.
You may report suspected misuse or safety violations through the existing text-report feature or the contact email in Section 29. Upon receiving a good-faith report or otherwise obtaining actual knowledge of a violation, we will promptly investigate and take reasonable steps within our control to stop or mitigate it, including restricting relevant content or services and addressing repeated violations where appropriate. This does not give us remote access to media retained only on your device or guarantee that fully offline content can be remotely removed.
The App currently does not provide user accounts, sign-in, cloud synchronization, user profiles, or cross-device cloud backup.
Except when you choose reports, feedback, connected mode, a custom server, model downloads, app-store functions, or another feature requiring a network connection, ordinary use data is primarily processed on your device.
You are responsible for managing, deleting, backing up, and protecting data, models, settings, prompts, source images, video and audio, text outputs, generated images and videos with audio, generated music, transcription outputs, caches, and exported files on your device. Uninstalling the App, clearing App data, or resetting the device may make data in the App's private storage unrecoverable, while files exported to a photo library, shared storage, a backup service, or another app may not be deleted with the App.
The App's features, including on-device video generation with audio, are free to use. There is no paid feature unlock, paid subscription entitlement, or generation-credit charge. The App may offer voluntary support payments through in-app purchases solely to support maintenance, development, and improvement; such support does not change feature access. Device, compatibility, region, and model-license conditions still apply, and third-party connectivity or services may have their own charges.
You understand and agree that:
support is entirely voluntary;
support does not unlock additional features, models, content, privileges, badges, priority support, subscriptions, or other digital goods;
the App's ordinary features are available on the same terms whether or not you provide support;
support is not a charitable donation, tax-deductible contribution, investment, pre-order, subscription, promise of future functionality, or guarantee of a service period;
payments, receipts, taxes, refunds, and transaction records may be processed by Google Play or the app store, payment service, or platform you use; and
refunds, cancellations, transaction disputes, and payment issues are governed by platform rules and applicable law.
We value your privacy. Please review the Gestura Privacy Policy for information about how the App collects, uses, retains, transmits, and deletes data.
These Terms and the Gestura Privacy Policy together govern important aspects of your use of the App. If they conflict on a data-processing matter, the Privacy Policy controls for personal data and privacy, while these Terms control for service use, licensing, models, outputs, liability, and dispute resolution, except where mandatory law requires otherwise.
The App and its non-third-party code, interface, design, icons, text, features, structure, trademarks, logos, and other content are owned by or licensed to us or the applicable rights holders and are protected by copyright, trademark, and other intellectual-property laws.
Downloading, installing, or using the App does not transfer ownership of the App or any part of it to you. Third-party names, trademarks, project names, model names, service names, and logos displayed or used in the App belong to their respective owners and are used only for identification, description, or compatibility purposes.
In addition to other restrictions in these Terms, you may not use the App to:
violate any applicable law, regulation, order, sanction, export control, platform policy, third-party service term, or model license;
infringe another person's copyright, trademark, trade secret, privacy, publicity, personality, personal-data, or other rights;
process, input, transmit, generate, or distribute another person's personal data, intimate imagery, confidential information, or protected content without authorization;
generate, process, or distribute unlawful, fraudulent, defamatory, harassing, hateful, discriminatory, violent, exploitative, malicious, or otherwise improper content;
create child sexual abuse material, sexualized content involving minors, non-consensual intimate imagery, or other sexual content prohibited by law;
impersonate, defraud, fabricate evidence, create materially harmful deceptive synthetic media, or falsely imply another person's endorsement, consent, or participation;
obtain, access, intercept, damage, or interfere with another person's data, system, account, server, API, model service, or device;
distribute malware, viruses, phishing material, or spam, or conduct cyberattacks, exploit vulnerabilities, perform unauthorized security testing, or otherwise harm information security;
circumvent safety measures, content restrictions, territorial restrictions, license checks, payment mechanisms, or rate limits, including by tampering with or replacing App or model files, cracking protections, or making a false region declaration; abuse models or APIs; or conduct unauthorized automation;
download, run, modify, redistribute, or commercially exploit a model, file, image, video, audio, text, or other content you are not authorized to use;
use the App in a manner that may damage, interfere with, disable, or overburden the App, a third-party service, a network, or infrastructure; or
use the App or outputs for a prohibited high-risk purpose or without appropriate human and professional review.
The App is not designed, validated, or certified for medical diagnosis, legal advice, financial or investment decisions, credit assessment, insurance underwriting, employment decisions, educational admissions, law enforcement, immigration, weapons control, critical infrastructure, emergency response, or other safety-critical uses that may materially affect a person, property, rights, or significant interests.
You may not use an AI output as the sole basis for such a significant decision. Any use that is otherwise permitted requires independent verification, appropriate testing, human oversight, and a final decision by a qualified and accountable professional, together with compliance with applicable laws and professional standards. Human review does not authorize a use prohibited by a model license or its acceptable-use policy.
We may update, modify, limit, suspend, or discontinue all or part of the App from time to time, including to fix errors, improve performance, adjust AI features, add or remove on-device image or video-generation capabilities, update model support, change model sources, address device compatibility, or respond to legal, platform, security, technical, or operational factors.
We do not guarantee that the App, a feature, a model, or a download source will remain available permanently, operate without interruption or error, be completely secure, or be compatible with every device, system version, model, API, server, model format, or third-party service. An update may change outputs, settings, performance, storage requirements, or compatibility.
You may stop using the App and remove it from your device at any time. You remain responsible for data, images, videos, and audio exported or stored elsewhere.
If we reasonably believe that you have violated these Terms, applicable law, third-party rights, third-party service terms, or a model license, or that your use creates a security, legal, operational, or technical risk, we may, to the extent permitted by law, restrict, suspend, or terminate your use of the App or a feature. We may be technically unable to remotely restrict a fully offline feature, but that does not affect your obligations under these Terms.
Provisions concerning intellectual property, user responsibilities, third-party and model licenses, AI outputs, disclaimers, limitations of liability, indemnification, governing law, and dispute resolution survive termination or cessation of use.
TO THE MAXIMUM EXTENT PERMITTED BY LAW, THE APP IS PROVIDED "AS IS" AND "AS AVAILABLE." WE DISCLAIM ALL EXPRESS, IMPLIED, STATUTORY, AND OTHER WARRANTIES, INCLUDING WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE, NON-INFRINGEMENT, AND WARRANTIES ARISING FROM COURSE OF DEALING.
Without limiting the foregoing, we do not warrant that:
the App will be uninterrupted, error-free, secure, stable, or completely reliable;
AI-generated text, images, videos, audio tracks, music, or transcription outputs will be accurate, complete, authentic, original, unique, reproducible, lawful, appropriate, or suitable for your needs;
generated images, videos, audio tracks, music, or transcription outputs will be free of defects, bias, improper content, similarities to third-party content, or copyright, trademark, likeness, voice, privacy, recording, performance, or other rights disputes;
on-device image, video, or music generation or music transcription will complete on your device within a particular time, resolution, duration, frame rate, audio quality, synchronization, or performance level, or without heat, battery drain, storage use, crashes, or data loss;
the App will be compatible with every device, system, server, API, model, model format, or third-party service;
model lists, license information, model files, download links, safety components, or third-party services will always be accurate, complete, available, secure, or unmodified; or
a third-party model or output may lawfully be used, commercially exploited, modified, redistributed, protected, or used without infringing third-party rights.
You are responsible for backing up important data and assume the risks of using the App, on-device image, video, and music generation and music transcription, AI features, third-party models, custom APIs, and custom servers. Warranties that cannot lawfully be excluded are not affected by this Section.
TO THE MAXIMUM EXTENT PERMITTED BY LAW, WE WILL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, EXEMPLARY, OR CONSEQUENTIAL DAMAGES, OR FOR LOSS OF PROFITS, REVENUE, DATA, GOODWILL, BUSINESS OPPORTUNITIES, MODEL FILES, PROMPTS, SOURCE IMAGES, VIDEO OR AUDIO, TEXT OUTPUTS, OR GENERATED IMAGES, VIDEOS OR AUDIO TRACKS, ARISING OUT OF OR RELATING TO THE USE OF OR INABILITY TO USE THE APP, EVEN IF WE HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
If applicable law does not permit a complete exclusion of liability, our liability will be limited to the minimum extent permitted by law. To the extent permitted by applicable law, our total aggregate liability for all claims relating to the App will not exceed the amount you actually paid directly to us for use of the App during the twelve months immediately preceding the event giving rise to the claim. If you paid no amount, we will have no monetary liability.
These exclusions and limitations do not apply to willful misconduct, gross negligence, personal injury that cannot be excluded by law, mandatory consumer-protection obligations, or any other liability that cannot lawfully be excluded or limited. Some jurisdictions do not permit certain exclusions or limitations, so they may not fully apply to you.
To the extent permitted by law, you agree to indemnify and hold us harmless from third-party claims, investigations, losses, liabilities, damages, costs, and reasonable expenses arising from your violation of these Terms or law, infringement of third-party rights, misuse of the App, use of a custom server or API, download or use of a third-party model, input or generation of improper content, or use, publication, distribution, licensing, or commercial exploitation of an AI output.
This Section does not limit non-waivable rights under applicable consumer law and does not require you to indemnify us for conduct for which we are legally responsible.
We may update these Terms from time to time. If a change is material, we will provide reasonable notice, such as by updating the effective date or last-updated date in the App, on an app-store page, on the Terms page, or in another reasonable location.
Revised Terms take effect on the stated effective date. Your continued use of the App after an update constitutes acceptance of the revised Terms, except where applicable law requires separate express consent. If you do not agree to a revision, stop using and remove the App.
These Terms and any dispute concerning their interpretation or validity are governed by the laws of the Republic of China (Taiwan), without regard to conflict-of-laws rules.
The parties will first attempt in good faith to resolve any dispute arising out of or relating to these Terms or the App through discussion. If no resolution is reached, the parties agree, to the extent permitted by applicable law, that the Taiwan Taipei District Court will have jurisdiction as the court of first instance.
If you are a consumer in the European Union, European Economic Area, United Kingdom, or another jurisdiction, these Terms do not exclude or limit mandatory consumer rights, remedies, or jurisdictional protections under the laws of your place of residence. The European Union Online Dispute Resolution (ODR) Platform was discontinued on July 20, 2025, so these Terms do not include a link to the discontinued platform.
This Section governs disputes under the Gestura Terms. It does not amend an independent third-party license's own governing-law or jurisdiction provisions, including those in the MiniMax H3 agreement.
These Terms may be provided in Traditional Chinese, English, or other languages. If there is an inconsistency or ambiguity among language versions, the Traditional Chinese version controls unless mandatory applicable law provides otherwise. A translation is provided for convenience and must not be interpreted to expand or restrict either party's rights or obligations.
The language-priority rule above applies only to the Gestura Terms. It does not replace or amend the original MiniMax H3 agreement, Apache-2.0 license, or other third-party license text. Any description or translation of those licenses in these Terms is for explanation only; the applicable original license text remains controlling for that license.
If any provision of these Terms is held invalid, unlawful, or unenforceable, the remaining provisions remain effective. The invalid or unenforceable provision will, to the extent permitted by law, be interpreted or replaced in the valid manner that most closely reflects its original purpose.
Our failure to immediately enforce a right under these Terms is not a waiver. Any waiver must be in writing and applies only to the specific instance for which it is given.
You may not assign your rights or obligations under these Terms without our prior written consent. We may assign our rights or obligations to the extent permitted by law in connection with a reorganization, service transfer, merger, acquisition, or transfer of assets.
These Terms, the Gestura Privacy Policy, and applicable third-party or open-source licenses constitute the complete agreement between you and us concerning use of the App. Headings are for convenience only and do not affect interpretation.
For questions about these Terms, the App, the Privacy Policy, data rights, model licenses, generated images, videos or audio, or other usage rights, contact us at:
Developer: Max Chao (趙士翔)
Email: zawshishawn@gmail.com
Location: Taipei, Taiwan
生效日期:2026 年 10 月 14 日
最後更新日期:2026 年 10 月 14 日
服務提供者:趙士翔(Max Chao)
所在地:台灣台北
聯絡信箱:zawshishawn@gmail.com
歡迎使用 Gestura(以下稱「本 App」)。本服務條款(以下稱「本條款」)規範您下載、安裝、存取、瀏覽或使用本 App 時,您與個人開發者趙士翔(Max Chao)(以下稱「我們」)之間的權利與義務。
當您下載、安裝、開啟或使用本 App,即表示您已閱讀、理解並同意受本條款拘束。若您不同意本條款,請勿下載、安裝或使用本 App。若適用法律要求另行取得明確同意,我們將依該法律要求辦理。
Gestura 是一款行動應用程式,提供文字與圖像處理、AI 文字輸出、本地圖片生成、依支援的文字、圖片、影片與音訊輸入在本地生成有聲影片、本地音樂生成與轉譜、模型清單瀏覽與下載、離線 AI 執行,以及連接使用者自訂伺服器或 API 端點等功能。實際可用功能可能因 App 版本、裝置、作業系統、地區、模型、硬體能力或第三方服務而異。
本 App 可能包含下列模式或功能:
離線模式:文字、圖像、影片、音訊、模型推論及其他支援的 AI 處理可於您的裝置本地端執行。本 App 不會僅為執行該等本地處理,而將本地輸入、提示詞或輸出傳送給我們或外部伺服器。
本地生成與轉譜功能:您可使用儲存在裝置上的相容模型生成圖片、有聲影片或音樂,或進行轉譜。MiniMax H3 相關影片功能可接受支援的文字、圖片、影片及音訊輸入,並在您的手機本地完成生成;此功能不是 MiniMax 或 Hugging Face 的雲端推論服務。模型下載及您另行選擇功能的網路連線,與本地生成及轉譜運算有所區別。
連線模式/自訂伺服器模式:您可自行設定並連接至 Ollama 伺服器、其他自訂 API 端點或第三方服務。於此模式下,您提交的文字、圖像、提示詞、上下文或其他請求內容可能會傳送至您指定的伺服器或 API,以產生文字、圖片或其他 AI 輸出。
模型清單與模型下載功能:本 App 可能透過 Hugging Face API 或其他公開來源取得模型清單、名稱、描述、作者、授權資訊、檔案資訊或下載連結。使用 H3 相關功能時,您會透過本 App 從 Hugging Face 下載由開發者提供的量化模型及相關元件,並應遵守第 11 條。
回報/意見反應功能:您可提交文字回報,並一併傳送 App 版本、裝置型號、Android 系統版本及系統語言設定。僅在您按下提交時才會傳送這些資料。App 內回報不接受圖片、影片、音訊或縮圖附件,且本 App 不會自動上傳生成、轉譜紀錄或診斷資料。
自願贊助功能:本 App 可能提供 App 內購買形式的自願贊助。贊助不會解鎖任何額外功能、內容、特殊權限、模型、徽章、優先支援或其他數位商品。
本 App 的資料處理方式請另參閱《Gestura 隱私政策》。
您必須符合您所在地法律允許使用本 App 及同意本條款的年齡與行為能力要求。
本 App 並非針對未滿 13 歲之兒童設計。若您未滿 13 歲,請勿使用本 App。若您所在地法律規定較高的最低年齡、家長同意或監護人同意門檻,應以該規定為準。於歐盟或歐洲經濟區,兒童就資訊社會服務之個人資料處理自行同意的年齡門檻,可能依會員國法律介於 13 至 16 歲。
若您尚未具備完整締約能力,您必須在父母、法定監護人或其他依法有權之人的同意及監督下使用本 App,且該人應代表您閱讀並同意本條款。
在您遵守本條款及適用法律的前提下,我們授予您一項有限、非專屬、不可轉讓、不可再授權、可撤回的授權,僅供您在個人或依法管理的裝置上下載、安裝及使用本 App。
除非適用法律明文允許、開源授權另有規定,或我們另行書面同意,您不得:
複製、修改、翻譯、改作、散布、出售、出租、轉讓或再授權本 App;
對本 App 進行反向工程、反編譯、拆解,或試圖取得非公開原始碼、演算法或技術秘密;
移除、隱藏或修改本 App 中的著作權、商標、開源授權、歸屬標示或其他權利聲明;
將本 App 用於建立與本 App 實質競爭的產品或服務,但適用法律或開源授權明確允許者不在此限;
以可能損害、干擾、癱瘓或過度負荷本 App、第三方 API、伺服器、模型來源或相關系統的方式使用本 App。
本 App 中使用的第三方或開源元件仍依其各自授權條款提供。本條款不取代、限制或削弱該等授權所賦予或限制的權利。
您應自行對透過本 App 輸入、提交、處理、傳送、下載、生成、儲存、匯出、發布或使用的內容負責,包括文字、提示詞、參考圖片、來源影片與音訊、人物肖像與聲音、錄音、音樂、歌詞、樂譜、表演、AI 文字輸出、生成圖片、影片及其音軌、生成音樂、轉譜結果、模型檔案、回報內容與意見反應。
您聲明並保證,您對相關內容具有必要的所有權、授權、同意或其他合法依據,且您的使用不會違反法律、本條款、第三方權利、模型授權、平台政策或第三方服務條款。
您應採取合理措施保護裝置、模型、API 金鑰、伺服器位址、輸入內容、輸出內容及其他資料,並自行判斷是否適合將敏感、機密、個人或受保護資訊輸入本 App或傳送至第三方服務。
本 App 可處理所選功能支援的文字、圖片、影片及音訊輸入,並依模型、設定及運算環境產生 AI 文字、圖片、有聲影片、音樂或轉譜結果。AI 輸出是機率性結果,不代表我們、模型作者或任何第三方的陳述、認可或保證。
您理解並同意:
AI 文字、圖片、影片、音軌、音樂或轉譜結果可能不準確、不完整、過時、具偏見、令人不悅、不安全、與提示詞不符或不適合特定用途;
輸出可能包含虛構資訊或話語、失真的人物或物件、錯誤文字、視覺或音訊瑕疵、時間連續性問題或影音不同步,亦可能與既有作品、聲音、錄音、音樂、表演、人物、商標、角色、風格或其他內容相似;
相同或相似的提示詞可能產生不同結果,不同使用者亦可能取得相同或相似的輸出;輸出不保證具有唯一性;
AI 輸出不應被視為事實、真實事件紀錄、專業意見、法律意見、醫療建議、財務建議、投資建議或其他需要專業判斷的建議;
您應在使用、發布、分享、出售、授權、散布或依賴輸出前,自行審查、判斷、驗證並取得必要權利或同意;
因您使用、依賴、修改、發布、散布或商業利用 AI 輸出所生之風險,由您自行承擔。
我們不保證 AI 輸出的正確性、完整性、原創性、合法性、可著作權性、適用性、安全性、可重現性、可商業使用性或不侵害第三方權利。轉譜結果可能出現音高、節奏或音符辨識錯誤,使用前應自行校對。
本地圖片、影片、音樂生成與轉譜功能使用您裝置上的模型、支援的輸入、提示詞、設定及運算資源。H3 相關影片功能會在本地處理文字、圖片、影片及音訊,以生成有聲影片。本地生成與轉譜不會將該等輸入或輸出上傳給我們、MiniMax、Hugging Face 或其他外部伺服器。模型下載是獨立的網路操作,不是上傳您的生成或轉譜內容。
您理解並同意:
模型下載、App 更新、應用商店功能、授權資訊查詢、您設定的自訂端點、回報或分享等功能仍可能需要網路連線;
本地圖片、影片、音樂生成與轉譜可能大量使用 CPU、GPU、NPU、記憶體、儲存空間、電量及其他裝置資源,並可能造成發熱、耗電、降頻、背景程序終止或其他效能影響;
生成與轉譜速度、影片與音樂長度、解析度、幀率、影像與聲音品質、影音同步、轉譜品質、穩定性及可用模型,取決於裝置能力、作業系統、模型格式、模型大小、可用記憶體、儲存空間及其他技術條件;
提示詞、輸入、設定、預覽、快取、暫存檔及輸出,可能依本 App 的設計與您的操作儲存在裝置上。生成影片預設保存在 App 內,不會自動存入相簿;您按下下載或匯出操作後,影片才會儲存至相簿。匯出檔案在您移除本 App 後仍可能保留;
生成檔案及轉譜結果可能包含時間、尺寸、長度、格式或其他技術中繼資料。在分享前,您應考量隱私,以及依法或依授權必須保留的署名、AI 生成揭露、浮水印或出處資訊;
內容過濾、安全檢查、提示或警告,均無法保證攔截所有違規內容。其存在或未攔截內容,不代表輸出合法、適當、經我們審核或獲得授權。您不得藉由竄改檔案、破解防護或其他繞過安全措施的方式生成違規內容。
您應自行管理、備份及刪除本地生成內容與轉譜結果,並在裝置遺失、共用、送修、出售或報廢前採取適當的資料保護措施。
您保留您原本對提示詞、來源圖片、影片、音訊及其他輸入內容所擁有的權利。對於完全在您裝置本地產生且未主動提交給我們的圖片、影片及其音軌、音樂或轉譜結果,我們不主張取得其所有權。於我們依法可能對該等輸出享有任何權利的範圍內,我們不主張該等權利;但此表述不代表您必然取得著作權、其他專有權或不受限制的使用權。
您對生成圖片、影片、音軌、音樂與轉譜結果的使用範圍可能受到下列因素影響:
提示詞、來源圖片、影片、音訊、錄音、音樂、歌詞、樂譜、表演、參考內容或其他輸入本身的權利狀態;
所使用模型、模型權重、附加元件、資料或相關工具的授權條款及可接受使用政策;
自訂伺服器、第三方 API、應用商店或其他平台的條款;
適用的著作權、商標、肖像權、人格權、隱私、個人資料、營業秘密、消費者保護、廣告、選舉、出口管制或其他法律;
輸出是否與第三方作品、角色、品牌、人物、產品、地點或其他受保護內容相同或近似。
您應自行進行必要的權利查核、相似性檢查、授權確認及法律審查,包括聲音、錄音、音樂、歌詞、樂譜與表演所需的許可。我們不保證生成圖片、影片、音軌、音樂或轉譜結果可登記、取得著作權保護、商業使用、再授權、再散布、作為商標使用,或不受第三方權利主張。轉譜不會使您取得原音樂、錄音或樂譜的權利,亦不取代使用或散布轉譜結果所需的授權或其他合法依據。
若模型授權、法律、平台政策或交易對象要求揭露內容為 AI 生成、保留來源資訊、標示模型,或維持內容憑證、浮水印或其他出處資訊,您應遵守該等要求,不得以違法或誤導方式移除、偽造或隱匿。H3 相關內容亦適用第 11 條的特別規定。
若您輸入、生成、編輯或發布包含可識別人物、臉部、身體、聲音、話語、姓名、身分特徵、私密情境或其他人格資訊的圖片、影片或音訊,您應確認具有適用法律所要求的權利、同意或其他合法依據。持有或提供某人的照片、影片或錄音,本身並不代表該人已同意被合成呈現或模仿。
您不得使用本 App:
生成、修改、持有或散布任何涉及兒童性剝削、兒童性虐待內容,或以未成年人為性化對象的內容;
製作或散布未經當事人同意的私密影像、性影像、裸露合成影像,或以此威脅、勒索、羞辱或騷擾他人;
冒充他人、偽造身分、詐騙、進行未經授權的代言,或以足以造成實質損害的方式誤導他人對內容來源、真實性、人物言行、事件或關係的認知;
捏造證據、偽造官方文件或以欺騙方式影響司法、行政、就業、教育、金融、保險、選舉或其他重大程序;
侵害他人的肖像權、人格權、隱私權、名譽、個人資料權利或其他合法權益;
以仇恨、歧視、暴力、騷擾、跟蹤或其他非法方式針對個人或群體。
在適用法律、模型授權、平台政策、專業規範或避免合理誤解所必要的情況下,您應清楚揭露內容係由 AI 生成或經 AI 修改。本 App 產生的圖片、影片及音訊,不得被視為真實人物、話語、表演、事件、文件或場景的可靠證明。
於離線模式下,本 App 會在您的裝置本地端處理支援的文字、圖像、影片、音訊、模型推論、圖片生成、有聲影片生成、音樂生成、轉譜或其他 AI 運算。您應自行確保裝置具備足夠效能、記憶體、儲存空間、電量、安全設定及相容環境。
我們不保證離線模式或本地圖片、影片、音樂生成或轉譜在所有裝置、Android 版本、語言、模型格式、模型大小、硬體條件或系統環境下均可正常運作。裝置製造商、作業系統或其他 App 的限制亦可能影響功能。
您應自行備份重要資料。若因裝置故障、模型錯誤、系統更新、儲存空間不足、操作錯誤、檔案損壞、權限變更或其他原因造成資料、圖片、影片或音訊遺失、模型損壞或 App 無法使用,我們僅在法律不得排除的範圍內負責。
當您使用連線模式並自行連接至 Ollama 伺服器、自訂 API 端點或其他第三方服務時,您理解並同意:
您提交的文字、圖像、提示詞、上下文、設定、來源圖片或其他請求內容可能傳送至您指定的伺服器或 API,並由該服務處理或保存;
我們不控制您連接的伺服器、API、模型、第三方服務、其安全措施或資料處理方式;
您應自行確認端點位址、憑證、API 金鑰、加密、安全性、隱私政策、服務條款、資料保留、內容政策、費用、速率限制及可用性;
您不得使用未經授權的帳號、憑證、API 金鑰、運算資源或伺服器,並應防止金鑰或連線資訊外洩;
因您連接、設定或使用自訂伺服器、API 或第三方服務而產生的資料外洩、未授權存取、服務中斷、費用、錯誤輸出、內容限制或其他損害,由您自行承擔;
我們不對任何第三方伺服器、API、模型、服務、內容、付款或其行為負責。
第 6 條所述的 H3 本地影片功能,不會使用本連線模式執行生成。僅設定自訂端點,不會使該功能變成雲端影片生成。
Ollama、Hugging Face、MiniMax、MiniMax H3、llama.cpp 及其他第三方名稱、商標、服務名稱或專案名稱均屬其各自權利人所有。除非另有明示,本 App 與該等第三方服務、專案或權利人並無代理、合夥、僱傭、授權經銷、背書或從屬關係。
本 App 可能使用 Hugging Face API 或其他公開來源取得文字模型、圖片生成模型、有聲影片生成模型、音樂生成模型、轉譜模型及相關元件的清單、描述、作者資訊、授權標示、檔案資訊或下載連結。相關元件可能包括模型權重、量化檔案、差分檔案、文字或多模態編碼器、影片或音訊解碼器、VAE、LoRA、介面卡、詞彙表、安全元件或其他檔案。
您理解並同意:
模型、模型權重、資料、程式碼、模型卡、說明文件及其他內容可能由不同作者、組織或權利人提供;
不同模型及元件可能適用不同授權、使用限制、可接受使用政策、研究用途限制、非商業限制、署名要求、散布條件或輸出相關條件;
模型是否可商業使用、修改、再散布、嵌入產品、提供服務或用於特定類型內容,應以模型頁面、模型卡、授權文件及權利人條款為準;
本 App 顯示模型或提供下載入口,不代表我們擁有、審查、推薦、保證該模型安全或合法,亦不代表我們授權您以任何特定方式使用;
下載、執行、修改、再散布或商業使用前,您應自行閱讀並遵守所有適用條款;若授權不明或與用途不相容,您不應使用該模型或元件;
您不得移除、隱藏或修改模型、模型卡、檔案、授權文件或第三方內容中的著作權、授權、來源、限制或其他權利聲明;
第三方模型可能含有錯誤、惡意程式碼、不安全內容、偏見、技術缺陷或不相容檔案,並可能占用大量頻寬、儲存空間或運算資源;您應自行評估來源與風險。
對於可歸責於您下載、使用、修改、再散布、提供服務或商業利用第三方模型或元件所生的法律爭議、權利主張、損害、費用或責任,您應在法律允許的範圍內負責。此規定不排除我們作為 App 提供者或修改後模型檔案散布者自身應負的義務與責任,亦不將該等義務與責任轉嫁給您。
H3 相關影片功能使用由開發者提供的量化模型及相關元件,並透過本 App 從 Hugging Face 下載。
Gestura H3 模型套件的概括修改聲明指出,適用 files.json 清單所列的模型資料檔案(model payloads),均由 Gestura 透過格式轉換、量化或差分封裝進行修改,並非未經修改的上游原始發布版本。此聲明針對清單所列的模型資料檔案;其他檔案與元件仍依其各自適用的權利聲明處理。
原始模型為 MiniMax H3。模型套件及相關元件適用的來源與權利聲明,請參閱本 App 共用授權區。相關聲明不暗示任何上游作者對 Gestura 或其修改提供背書。
就本 App 共用授權區標示為 Apache-2.0 的另行授權第三方貢獻,該授權僅適用於相應授權人實際授權的權利與內容,不會將 MiniMax H3 重新授權,亦不移除 H3 協議對 H3 相關作品及衍生模型的條件。適用的 Apache-2.0 授權副本、修改標示及歸屬聲明要求亦須遵守。
所有適用的上游與元件授權所定的權利與限制仍然有效,不會因量化、轉檔、封裝或免費提供而消失。模型庫名稱、下載入口或摘要,不取代適用的授權檔案與權利聲明。
本 App 內附《MiniMax H3 Community License Agreement》完整協議及附件 A(Exhibit A,可接受使用政策)。在標準模型下載流程中,本 App 會先顯示上述文件,您必須勾選並同意,才能繼續下載相關模型。
下載或使用 H3 相關功能前,您必須閱讀並同意遵守《MiniMax H3 Community License Agreement》,包括其第 V 節及附件 A(可接受使用政策)。該等使用限制納入本條款,並於您存取或使用該功能時對您具有拘束力。以下摘要並非全部限制;H3 相關作品以完整協議為準,本條款不額外授予模型或輸出的使用權。
https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE
依日期為 2026 年 8 月 2 日的授權,排除地區為歐盟、英國、韓國及美國。地域限制涵蓋在該等地區使用、重製、修改、散布及展示 H3 相關作品與其輸出。向公開環境發布 H3 生成內容時,必須清楚且顯著揭露為 AI 生成。不得使用 H3 相關作品或輸出改善 H3 及其衍生模型以外的其他 AI 模型。
在開放相關模型下載前,本 App 會詢問您是否位於限制地區;若您回答位於限制地區,本 App 不會開放該模型下載。地區聲明答案僅在您的手機本地處理,不會因該聲明流程而傳送給我們或第三方。
您應如實回答,不得透過虛偽聲明或竄改檔案繞過限制。通過下載檢查或先前已下載模型,不代表後續模型使用或輸出散布已獲授權。上述 H3 條件不取代其他模型各自適用的授權。
本 App 共用授權區的模型來源文件內附 H3 的 NOTICE 聲明內容,以及 Gestura 轉換資產屬修改版本的概括聲明。您可於該區查閱;目前的下載前對話框未載入該模型來源文件。下述授權副本、NOTICE 文字檔及修改標示要求,不因文件顯示位置而改變。
對於我們散布或提供的 H3 相關作品,我們仍應依 H3 協議第 III 節向接收者提供完整協議、使修改檔案帶有顯著修改聲明,並於非託管服務形式的散布中隨附必要的 NOTICE 文字檔。該必要聲明如下:
MiniMax H3 is licensed under the MiniMax H3 Community License Agreement, Copyright © 2026 MiniMax. All Rights Reserved.
對於另行授權的元件,亦須保留適用的授權及上游權利聲明。模型卡標籤、來源網址或本摘要,不取代應提供的授權副本或 NOTICE 檔案。若您進一步散布 H3 相關作品,其散布要求及整合式終端產品接收者可能適用的例外,依完整 H3 協議辦理;該協議第 V 節及附件 A 仍繼續適用。
本條款不取代我們依 H3 協議第 V.5 節應負的提供者責任,包括實施、維護、測試及定期檢視合理且相稱的防護措施、維持可使用的回報機制,並在控制範圍內調查及合理處理違規情形。本地運作及使用者責任,不會免除該等提供者義務。
本 App 可能使用第三方軟體、開源元件、SDK、API、模型執行框架、模型來源、應用商店或平台服務。該等元件及服務仍依其各自授權條款、第三方條款或平台政策提供;本條款不取代、限制或變更該等條款。
本 App 使用之開源與第三方元件清單、著作權聲明、授權文字、NOTICE 檔案及免責聲明,請參閱 App 內或相關頁面的「開源授權」頁面。若第三方或開源授權條款與本條款有所不同,就該元件、SDK、API、模型或第三方服務本身,應優先適用其各自條款。
第 3 條的 App 授權適用於 Gestura 軟體,不是 H3 相關作品或另行授權元件的替代授權。本條款不對 H3 修改版本施加違反 H3 協議的額外或不同條件。
本 App 提供純文字形式的「回報」、「意見反應」或類似功能。只有在您主動按下提交時,我們才會接收回報及隨附的裝置與系統資訊。本 App 不會自動上傳生成、轉譜紀錄或診斷資料;生成圖片、影片、音樂或進行轉譜本身,也不會使媒體、轉譜結果或縮圖傳送給我們。
App 內回報所傳送的資料限於:
您選擇提交的文字,例如意見、問題說明,或您刻意納入的 AI 文字與提示詞;
App 版本、裝置型號、Android 系統版本及系統語言設定,僅在您提交回報時隨文字一併傳送。
App 內回報不接受圖片、影片、音訊或縮圖附件。模型細節、設定或錯誤資訊,僅在您選擇於回報文字中描述時才會包含;此功能不會自動上傳其他診斷檔案。
您授予我們一項全球性、非專屬、免權利金、可儲存、使用、重製、分析及處理您主動提交文字的有限授權,目的僅限於診斷與修復錯誤、改善穩定性與效能及 AI 功能可靠性、提供語言支援、回應您的請求、處理回報的濫用情形及遵守法律義務。此授權不及於僅保存在您裝置上的媒體或其他內容。
請勿在回報中提交不必要的個人資料、私密細節、機密資訊、第三方秘密或您無權提供的內容。除非法律另有規定或您另行同意,我們不會將該等資料用於廣告、行銷、跨服務追蹤、廣告輪廓分析或出售給第三方。保存期間、刪除請求及其他資料權利依《Gestura 隱私政策》辦理。
您可透過既有純文字回報功能或第 29 條的聯絡信箱,回報疑似濫用或違反安全規範的情形。收到善意回報或以其他方式實際知悉違規後,我們會儘速調查,並在控制範圍內採取合理措施以停止或減輕違規情形,包括於適當情況下限制相關內容或服務及處理重複違規。此規定不使我們取得對僅保存在您裝置上媒體的遠端存取權,亦不保證可遠端移除完全離線的內容。
本 App 目前不提供使用者帳號、登入系統、雲端同步、使用者個人檔案或跨裝置雲端備份功能。
除非您主動使用回報、意見反應、連線模式、自訂伺服器、模型下載、應用商店或其他需網路連線的功能,一般使用資料主要於您的裝置本地端處理。
您應自行管理、刪除、備份及保護裝置上的資料、模型、設定、提示詞、來源圖片、影片與音訊、文字輸出、生成圖片與有聲影片、生成音樂、轉譜結果、快取及匯出檔案。刪除 App、清除資料或重置裝置可能使 App 私有儲存區中的資料無法復原;但您已匯出至相簿、共享儲存空間、備份服務或其他 App 的檔案可能不會一併刪除。
本 App 的功能(包括本地有聲影片生成)免費使用,沒有付費功能解鎖、付費訂閱權益或生成點數費用。本 App 可能提供 App 內購買形式的自願贊助,僅用於支持維護、開發與改善;贊助不影響功能存取。裝置、相容性、地區及模型授權條件仍然適用,第三方網路連線或服務亦可能另行收費。
您理解並同意:
贊助完全自願;
贊助不會解鎖任何額外功能、模型、內容、特殊權限、徽章、優先支援、訂閱服務或其他數位商品;
無論是否贊助,您均可在相同條件下使用本 App 的一般功能;
贊助不是公益捐款、慈善捐贈、投資、預購、訂閱、保證未來功能或保證服務期間;
付款、收據、稅費、退款與交易紀錄可能由 Google Play 或您使用的應用商店、付款服務或平台處理;
退款、取消、交易爭議及付款問題依平台規則及適用法律辦理。
我們重視您的隱私。本 App 如何收集、使用、保存、傳輸與刪除資料,請參閱《Gestura 隱私政策》。
本條款與《Gestura 隱私政策》共同構成您使用本 App 的重要規範。若兩者就資料處理事項有不一致之處,涉及個人資料或隱私保護者,以隱私政策為準;涉及服務使用、授權、模型、輸出、責任限制或爭議處理者,以本條款為準,但適用法律另有強制規定者從其規定。
本 App 及其非第三方部分之程式碼、介面、設計、圖示、文字、功能、結構、商標、標誌及其他內容,均由我們或相關權利人擁有或授權使用,並受著作權、商標權及其他智慧財產權法律保護。
您不會因下載、安裝或使用本 App 而取得本 App 或其任何部分的所有權。本 App 中顯示或使用的第三方名稱、商標、專案名稱、模型名稱、服務名稱或標誌均屬其各自權利人所有,僅為識別、說明或相容性目的使用。
除本條款其他限制外,您不得使用本 App 從事下列行為:
違反任何適用法律、法規、命令、制裁、出口管制、平台政策、第三方服務條款或模型授權;
侵害他人的著作權、商標權、營業秘密、隱私權、肖像權、人格權、個人資料權利或其他權利;
未經授權處理、輸入、傳送、生成或散布他人的個人資料、私密影像、機密資訊或受保護內容;
生成、處理或散布非法、詐欺、誹謗、騷擾、仇恨、歧視、暴力、剝削、惡意或其他不當內容;
製作兒童性虐待內容、性化未成年人內容、未經同意的私密影像或其他依法禁止的性內容;
冒充、詐騙、偽造證據、製作具實質危害的欺騙性合成媒體,或未經授權暗示他人代言、同意或參與;
取得、存取、攔截、破壞或干擾他人的資料、系統、帳號、伺服器、API、模型服務或裝置;
散布惡意程式、病毒、釣魚內容、垃圾訊息,進行網路攻擊、漏洞利用、未經授權的安全測試或其他危害資訊安全的行為;
藉由竄改或替換 App 或模型檔案、破解防護、虛偽地區聲明或其他方式,規避安全措施、內容限制、地區限制、授權檢查、付款機制或速率限制,濫用模型或 API,或進行未經授權的自動化操作;
下載、執行、修改、再散布或商業利用您無權使用的模型、檔案、圖片、影片、音訊、文字或其他內容;
以可能損害、干擾、癱瘓或過度負荷本 App、第三方服務、網路或基礎設施的方式使用本 App;
將本 App 或輸出用於法律禁止或未經適當人工與專業審查的高風險用途。
本 App 並非為醫療診斷、法律意見、金融或投資決策、信用評估、保險核保、就業決策、教育錄取、執法、移民、武器控制、關鍵基礎設施、緊急應變或其他可能對人身、財產、權利或重大利益造成重大影響的安全關鍵用途而設計、驗證或認證。
您不得將 AI 輸出作為上述重大決策的唯一依據。任何在其他規定下仍屬允許的使用,均應由具資格且負責任的專業人士進行獨立查核、必要測試、人類監督及最終決策,並遵守適用法律與專業規範。人工審查不會使模型授權或其可接受使用政策禁止的用途變成允許。
我們可能不定期更新、修改、限制、暫停或停止本 App 的全部或部分功能,包括修復錯誤、改善效能、調整 AI 功能、增加或移除本地圖片或影片生成能力、更新模型支援、調整模型來源、因裝置相容性進行變更,或因法律、平台政策、安全、技術及營運因素進行調整。
我們不保證本 App、特定功能、模型或下載來源會永久提供、不中斷、無錯誤、完全安全,或與所有裝置、系統版本、模型、API、伺服器、模型格式及第三方服務相容。更新可能改變輸出、設定、效能、儲存需求或相容性。
您可隨時停止使用本 App,並從裝置移除本 App。您仍應自行處理已匯出或保存在其他位置的資料、圖片、影片與音訊。
若我們合理認為您違反本條款、適用法律、第三方權利、第三方服務條款、模型授權,或您的使用可能造成安全、法律、營運或技術風險,我們得在法律允許範圍內限制、暫停或終止您使用本 App 或其部分功能。對於完全離線且技術上不受我們控制的功能,我們可能無法遠端限制,但不影響您依本條款應負的義務。
關於智慧財產權、使用者責任、第三方與模型授權、AI 輸出、免責聲明、責任限制、賠償、準據法及爭議處理的條款,在您停止使用本 App後仍繼續有效。
在法律允許的最大範圍內,本 App 係依「現狀」及「可用狀態」提供。我們不作任何明示、默示、法定或其他保證,包括適銷性、特定目的適用性、權利歸屬、不侵權及交易過程所生之保證。
在不限制前述規定的情況下,我們不保證:
本 App 將不中斷、無錯誤、安全、穩定或完全可靠;
AI 文字、圖片、影片、音軌、音樂或轉譜結果將正確、完整、真實、原創、唯一、可重現、合法、適當或符合您的需求;
生成圖片、影片、音軌、音樂或轉譜結果不會包含瑕疵、偏見、不當內容、與第三方內容的相似性,或不會引發著作權、商標、肖像、聲音、隱私、錄音、表演或其他權利爭議;
本地圖片、影片、音樂生成或轉譜在您的裝置上可於特定時間、解析度、長度、幀率、聲音品質、影音同步或效能下完成,或不會造成發熱、耗電、儲存空間占用、當機或資料遺失;
本 App 與所有裝置、系統、伺服器、API、模型、模型格式或第三方服務相容;
模型清單、授權資訊、模型檔案、下載連結、安全元件或第三方服務永遠正確、完整、可用、安全或未受修改;
任何第三方模型或輸出可合法使用、商業使用、修改、再散布、取得權利或不侵害第三方權利。
您應自行備份重要資料,並自行承擔使用本 App、本地圖片、影片、音樂生成與轉譜、AI 功能、第三方模型、自訂 API 或自訂伺服器的風險。適用法律不允許排除的保證不受本條影響。
在法律允許的最大範圍內,對於因使用或無法使用本 App 所產生或相關的任何間接、附帶、特殊、懲罰性、示範性或衍生性損害,或利潤、收入、資料、商譽、商業機會、模型檔案、提示詞、來源圖片、影片或音訊、文字輸出、生成圖片、影片或音軌之損失,我們不負責任,即使我們已被告知可能發生該等損害亦同。
若法律不允許完全排除責任,我們對您的責任應限於法律允許的最低範圍。在適用法律允許的範圍內,我們對所有與本 App 相關請求的累計總責任,以導致請求之事件發生前十二個月內您實際為使用本 App 支付給我們的金額為上限;若您未支付任何費用,則我們不承擔金錢賠償責任。
上述排除或限制不適用於因故意、重大過失、法律不得排除之人身損害、消費者保護強制規定或其他依法不得排除或限制的責任。部分司法管轄區不允許某些排除或限制,因此相關規定可能不完全適用於您。
若因您違反本條款或法律、侵害第三方權利、濫用本 App、使用自訂伺服器或 API、下載或使用第三方模型、輸入或生成不當內容,或使用、發布、散布、授權或商業利用 AI 輸出,而導致任何第三方請求、調查、損失、責任、損害、費用或合理支出,您同意在法律允許範圍內賠償並使我們免受損害。
本條不限制您依適用消費者保護法享有且不得放棄的權利,亦不要求您就依法應由我們負責的行為承擔賠償責任。
我們可能不定期更新本條款。若有重大變更,我們將透過合理方式通知您,例如於 App 內、應用商店頁面、條款頁面或其他合理位置更新生效日期或最後更新日期。
修訂後條款自標示之生效日期起生效。若您於條款更新後繼續使用本 App,即表示您同意修訂後條款,但適用法律要求另行取得明確同意者除外。若您不同意修訂內容,請停止使用並移除本 App。
本條款之解釋、效力及爭議處理,以中華民國(台灣)法律為準據法,但不適用其法律衝突規則。
因本條款或本 App 所生或與之相關的爭議,雙方應先以誠信方式協商解決。若協商不成,在適用法律允許的範圍內,雙方同意以臺灣臺北地方法院為第一審管轄法院。
若您為歐盟、歐洲經濟區、英國或其他司法管轄區的消費者,本條款不排除或限制您依居住地消費者保護法律享有的強制性權利、救濟或管轄保障。歐盟線上爭議解決(ODR)平台已於 2025 年 7 月 20 日停止運作,因此本條款不提供已停用平台的連結。
本條規範依 Gestura 服務條款所生的爭議,不修改獨立第三方授權自身的準據法或管轄約定,包括 MiniMax H3 協議的相關規定。
本條款可能提供繁體中文、英文或其他語言版本。若不同語言版本之間有任何不一致或歧義,除非適用法律另有強制規定,應以繁體中文版本為準。翻譯僅為便利理解,不應被解釋為擴張或限制任何一方的權利義務。
前述語言優先規則僅適用於 Gestura 服務條款,不取代或修改 MiniMax H3 協議、Apache-2.0 或其他第三方授權原文。本條款對該等授權的說明或翻譯僅為解釋用途;該授權本身仍以其適用原文為準。
若本條款任何部分被認定無效、違法或不可執行,不影響其他部分的效力。無效或不可執行部分應在法律允許範圍內,以最接近原意且有效的方式解釋或取代。
我們未立即執行本條款中的任何權利,不代表放棄該權利。任何放棄均須以書面為之,且僅適用於該次特定情況。
您不得未經我們事前書面同意轉讓本條款下的權利或義務。我們得在法律允許範圍內,因營運重組、服務移轉、合併、收購或資產移轉而轉讓本條款下的權利或義務。
本條款、《Gestura 隱私政策》及適用的第三方或開源授權,構成您與我們就本 App 使用事項的完整約定。各標題僅為閱讀便利,不影響條款解釋。
若您對本條款、本 App、隱私政策、資料權利、模型授權、生成圖片、影片或音訊,以及其他使用權利有任何問題,請透過以下方式聯絡我們:
開發者: 趙士翔(Max Chao)
Email: zawshishawn@gmail.com
所在地: 台灣台北