overview

These Terms of Service ("Terms") govern your access to and use of the services provided by Inference Shell Inc., a Delaware corporation ("inference.sh," "Company," "we," "us," or "our"). By accessing or using our platform at inference.sh, you ("Customer," "you," or "your") agree to be bound by these Terms.

inference.sh operates as an API platform and execution environment. We are the merchant of record for all transactions on the platform. When you purchase credits or services, you are transacting directly with Inference Shell Inc., not with third-party application creators or resource providers.

BY AGREEING TO THESE TERMS, YOU AND INFERENCE SHELL INC. AGREE TO RESOLVE DISPUTES THROUGH BINDING INDIVIDUAL ARBITRATION AND WAIVE ANY RIGHT TO A JURY TRIAL OR TO PARTICIPATE IN A CLASS ACTION. SEE "DISPUTE RESOLUTION; ARBITRATION" BELOW FOR DETAILS.

your data

inference.sh does not sell or commercially exploit Customer data without permission.

Customer retains ownership of all inputs and outputs. The Company will not sell, license, train on, or otherwise commercially exploit Customer content unless:

Customer provides explicit consent (e.g., for case studies, testimonials, or marketing), or
Customer elects to make content public (e.g., publishing applications, sharing outputs publicly, or contributing to community resources)

Content that is not made publicly available remains confidential and is processed solely to provide the Services. Content that Customer chooses to publish or share publicly will be available to other users under the terms specified by Customer.

The Company may generate anonymized, aggregated statistics (such as request volumes and error rates) that cannot be traced to any individual Customer. Such data is used to operate and improve the platform and does not include identifiable Customer content.

These commitments apply to all Customer data, including data submitted prior to this Terms update.

For complete details, see "Customer Input, Output Content, and Platform Usage Data" below.

account requirements

You must provide accurate, complete account information and maintain its security. You are responsible for all activity under your account. You must be at least 18 years old or the age of majority in your jurisdiction. We reserve the right to suspend or terminate accounts that violate these Terms.

prepaid credits

nature of credits

Credits on inference.sh are prepaid usage units, not money, stored value, or a financial instrument. Credits:

Represent prepaid access to platform services
Are non-transferable between users or accounts
Have no cash equivalent and cannot be redeemed for cash
Do not constitute a bank account, wallet, or stored value account
Are denominated in microcents (1 credit = 1/1,000,000 of a cent USD) for precision billing

purchasing credits

When you add funds to your account, you are purchasing prepaid usage credits from Inference Shell Inc. All purchases are processed through our payment provider (Stripe). Credits are available immediately upon successful payment. Upon purchase, funds become property of Inference Shell Inc. in exchange for prepaid services.

no fiduciary relationship

inference.sh does not act as a bank, escrow agent, or payment intermediary between users. We do not hold funds on behalf of users or facilitate payments between parties.

refunds and adjustments

Refunds for unused credits are provided at our sole discretion. We may offer refunds in cases of platform error, service unavailability, or other circumstances we deem appropriate. To request a refund, contact [email protected].

We reserve the right to reverse credits, adjust balances, or reclaim usage for fraud, abuse, chargebacks, or payment disputes. If an underlying transaction is reversed or disputed, any associated credits or earnings may be adjusted accordingly.

credit expiration

Purchased credits expire 12 months from the date of purchase if not used. Promotional or free credits expire 90 days from issuance unless otherwise specified. Upon expiration, unused credits are forfeited and will not be refunded or restored. It is your responsibility to monitor your credit balance and usage.

late payment

If you fail to make any payment when due, we may charge interest on past due amounts at the rate of 1.5% per month (or the highest rate permitted by law, if lower), calculated daily and compounded monthly. You agree to reimburse us for all reasonable costs incurred in collecting late payments, including attorneys' fees. If payment remains outstanding for 30 days following written notice, we may suspend access to the Services until all past due amounts and interest have been paid.

pricing and billing

dynamic pricing

Pricing for services may vary based on:

Compute resources consumed (CPU, GPU, memory, storage)
External API costs and third-party resource pricing
Bandwidth and data transfer
Model inference costs
Real-time market conditions for resources

Prices are displayed before execution when possible. For dynamic workloads, estimated costs are shown with actual charges calculated upon completion. Pricing may change between execution start and completion due to upstream cost changes. Executions may continue to incur charges until explicitly stopped or finished.

usage-based billing authorization

By using the platform, you authorize us to automatically deduct credits from your prepaid balance when:

Jobs, tasks, or executions run
API calls are made
Resources are provisioned
Retries or recovery operations occur
Tools or external services are invoked on your behalf

Deductions occur in real-time as services are consumed.

pass-through costs

Some executions may incur costs from external providers (APIs, models, compute resources). These costs are passed through to you as part of our service pricing. We procure these resources on your behalf and bill you for the complete service—you are not transacting directly with upstream providers.

service execution

how execution works

When you run workloads on inference.sh, our platform may:

Execute code in sandboxed environments
Make API calls to external services
Provision compute resources dynamically
Invoke third-party models or tools
Process and return results

All execution occurs on our infrastructure or through providers we contract with directly.

automated resource procurement

Our platform may automatically acquire external resources to fulfill your requests. For example, if an API returns a payment-required response, we may complete that payment and charge your account for the resulting service. This is resource procurement on our behalf to deliver your requested execution—not a payment made on your behalf to a third party.

third-party services and outputs

inference.sh integrates with external infrastructure, APIs, models, and tools operated by independent third-party providers ("Third-Party Services"). These Third-Party Services are not under our control, and their availability, performance, pricing, policies, or outputs may change at any time.

You acknowledge that the functionality of the Services may depend on Third-Party Services and that such services may not operate without interruption or error. inference.sh does not guarantee the accuracy, legality, reliability, or performance of Third-Party Services or any outputs generated through them.

When workloads invoke Third-Party Services, Customer Input may be transmitted to those providers solely as necessary to perform the requested execution. Your use of Third-Party Services through inference.sh is at your own risk.

To the maximum extent permitted by law, inference.sh shall not be liable for damages, losses, or claims arising from Third-Party Services, including model outputs, API responses, infrastructure outages, or changes to upstream provider terms or pricing.

acceptable use

You agree not to use the Services to:

general prohibitions

Use the platform for illegal purposes or to violate any laws
Abuse, exploit, or attempt to bypass third-party services through our platform
Circumvent pricing, metering, rate limits, or usage controls
Attempt to access other users' accounts, data, or resources
Distribute malware, conduct attacks, or compromise system security
Generate content that violates third-party terms of service
Resell platform access without authorization
Use automated systems to abuse platform resources

harmful content

Generate, promote, or distribute content involving child sexual abuse material (CSAM) or the exploitation of minors
Create non-consensual intimate imagery or sexual content depicting real individuals
Produce extreme violence, gore, torture, or content glorifying harm
Generate content that promotes terrorism, violence, or hatred against protected groups
Create content designed to harass, threaten, stalk, or bully individuals

deception and manipulation

Generate or distribute misinformation, disinformation, or deceptive content intended to mislead
Create deepfakes or synthetic media depicting real individuals without their consent or legal right
Impersonate real individuals, brands, or organizations without authorization
Generate political advertisements, propaganda, or content designed to manipulate elections or public opinion
Represent AI-generated content as human-created in contexts where such disclosure is material
Create fake reviews, testimonials, or false online engagement

biometrics and surveillance

Conduct real-time biometric identification in public spaces for surveillance or law enforcement purposes
Build or contribute to facial recognition databases without explicit informed consent of depicted individuals
Evaluate, classify, or score individuals based on social behavior or personal traits in ways that lead to discriminatory treatment
Categorize individuals using biometric data to infer sensitive attributes including race, religion, sexual orientation, or political beliefs
Predict criminal behavior or assess risk based solely on personal traits or profiling

high-risk and restricted applications

Operate weapons systems, nuclear facilities, or critical infrastructure without appropriate safeguards
Provide medical, legal, financial, or other professional advice without appropriate qualifications and disclosures
Make fully automated decisions that produce legal effects or similarly significant impacts on individuals without human oversight
Develop chemical, biological, nuclear weapons, or conduct gain-of-function research on pathogens

likeness and intellectual property

Generate content that replicates the likeness of any real person, brand, character, or location unless you have independently obtained all necessary rights, releases, and clearances
Use outputs to develop, train, or improve competing AI models or services without authorization
Systematically extract, scrape, or harvest content, metadata, or training data from the Services

We may throttle, suspend, reject workloads, or terminate accounts at our discretion for violations or suspected abuse. Enforcement is at our sole discretion, and failure to enforce in some instances does not waive our right to enforce in others.

intellectual property

your content

You retain ownership of content you create or upload. By using the platform, you grant us a license to process, store, and transmit your content as necessary to provide services.

platform ip

All platform features, code, interfaces, and documentation are owned by Inference Shell Inc. and protected by intellectual property laws. You may not copy, modify, or reverse engineer platform components.

ai-generated outputs

Outputs generated through AI models on the platform may be subject to the terms of the underlying model providers. We make no claim of ownership over AI-generated outputs but also provide no guarantee of exclusivity or intellectual property rights in such outputs.

customer input, output content, and platform usage data

The Services allow you to submit prompts, files, code, media, and other materials ("Customer Input") and receive generated or processed results ("Output Content"). As between you and Inference Shell Inc., you own and will continue to own all rights, title, and interest, including all intellectual property rights, in and to your Customer Input and Output Content. Nothing in these Terms transfers any ownership in Customer Input or Output Content to inference.sh or any other party.

To the extent needed, inference.sh hereby grants you all right, title, and interest, if any, in and to Output Content, including your use of Output Content for commercial purposes such as sale or publication, subject to any applicable Model Terms.

You grant inference.sh a limited, worldwide, non-exclusive, non-sublicensable (except to Third-Party Services solely as necessary to provide the Services), royalty-free license to use, reproduce, store, host, transmit, process, and display Customer Input solely for the following purposes:

Providing the Services and generating Output Content you request
Operating, securing, and maintaining the platform infrastructure
Complying with legal obligations

We do not use your Customer Input or Output Content to train AI models, and we do not sublicense your content to third parties for training purposes.

Due to the nature of AI systems and automated execution environments, Output Content may be non-deterministic, non-unique, inaccurate, incomplete, or infringing. inference.sh makes no representation or warranty that Output Content will be original, lawful, accurate, or fit for any particular purpose. You are solely responsible for reviewing and evaluating Output Content before relying on or distributing it.

You represent and warrant that you have all necessary rights, licenses, and permissions to submit Customer Input and to use any Output Content generated through the Services. You are solely responsible for compliance with applicable laws and third-party rights related to your use of the Services, Customer Input, and Output Content.

platform usage data

inference.sh may generate and use anonymized, aggregated, or de-identified technical data derived from operation of the Services ("Platform Usage Data"). Platform Usage Data does not include identifiable Customer Input or Output Content. Examples include aggregate request volumes, error rates, latency metrics, and feature usage patterns.

inference.sh owns all rights in Platform Usage Data and may use it to operate, monitor, secure, analyze, and improve the platform and related technologies. Platform Usage Data may be retained indefinitely.

inference.sh does not sell or commercially exploit identifiable Customer Input or Output Content without your explicit prior consent.

service availability

no uptime guarantees

inference.sh is provided "as is" without guarantees of availability, uptime, or reliability. Services may be:

Interrupted for maintenance or updates
Subject to capacity constraints
Affected by third-party outages
Unavailable in certain regions

experimental features

Some features may be marked as experimental, beta, or preview. These features may change, break, or be discontinued without notice.

non-deterministic outputs

AI and agent-based workloads may produce non-deterministic results. The same inputs may yield different outputs across executions. We do not guarantee consistency, accuracy, or fitness for any particular purpose.

ai and model disclaimers

accuracy and reliability

THE SERVICES UTILIZE ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING TECHNOLOGIES THAT ARE INHERENTLY PROBABILISTIC AND EXPERIMENTAL. AI MODELS MAY GENERATE OUTPUTS THAT ARE INACCURATE, INCOMPLETE, MISLEADING, OR ENTIRELY FABRICATED ("HALLUCINATIONS"). inference.sh DOES NOT GUARANTEE THE ACCURACY, RELIABILITY, COMPLETENESS, OR APPROPRIATENESS OF ANY AI-GENERATED OUTPUT.

You acknowledge that AI technologies are novel and evolving, and that significant uncertainty exists regarding their operation. You are solely responsible for reviewing, validating, and determining the suitability of any Output before relying on or distributing it.

no professional advice

The Services and any Output do not constitute and are not intended to provide legal, medical, financial, tax, accounting, or other professional advice. You must make your own independent determinations and consult qualified professionals as appropriate before acting on any information or recommendations generated by the Services.

inappropriate or harmful outputs

The Services may be capable of generating content that is inappropriate, offensive, inaccurate, illegal in certain jurisdictions, or otherwise objectionable. inference.sh does not monitor all generated content and expressly disclaims responsibility for any such outputs. You are solely responsible for reviewing, filtering, and managing all Output generated through your use of the Services.

third-party model compliance

Third-Party Services, including AI models, may have been developed using training data or methods that could infringe third-party rights or may not comply with laws in all jurisdictions. inference.sh does not warrant that any Third-Party Services or their outputs are non-infringing, lawful, or suitable for any purpose. You are responsible for ensuring that your use of outputs does not infringe third-party intellectual property rights, including by human review where necessary.

limitation of liability

To the maximum extent permitted by law, Inference Shell Inc. shall not be liable for:

Indirect, incidental, special, consequential, or punitive damages
Loss of profits, data, or business opportunities
Service interruptions or failures
Actions of third-party providers
Costs exceeding the amount you paid in the 12 months preceding the claim

indemnification

You agree to indemnify and hold harmless Inference Shell Inc., its officers, directors, employees, and agents from any claims, damages, or expenses arising from your use of the platform or violation of these Terms.

modifications

We may modify these Terms at any time. Material changes will be communicated via email or platform notification. Continued use after changes constitutes acceptance of the modified Terms.

export controls and sanctions

You agree to comply with all applicable export control and trade sanctions laws and regulations, including those administered by the United States. You represent and warrant that you are not located in, organized in, or ordinarily resident in any country or region subject to comprehensive U.S. embargoes and that you are not listed on any U.S. government restricted party list. You may not use the Services to export, re-export, transfer, or provide access to the platform or any outputs in violation of applicable export control laws.

governing law

These Terms are governed by the laws of the State of Delaware, USA, without regard to conflict of law principles. Any disputes shall be resolved in the state or federal courts located in Delaware.

Nothing in these Terms limits mandatory consumer protections that apply in your local jurisdiction.

dispute resolution; arbitration

informal resolution first

Before initiating arbitration, you must first send a written notice describing the dispute to [email protected] (a "Dispute Notice"). The Dispute Notice must include your name, account information, a description of the nature and basis of the claim, and the specific relief sought. We will attempt to resolve the dispute informally for 30 days after receiving your Dispute Notice. You may not commence arbitration until this 30-day informal resolution period has concluded. The statute of limitations for any claim will be tolled during this period.

binding individual arbitration

To the fullest extent permitted by law, any dispute, claim, or controversy arising out of or relating to these Terms or the Services that is not resolved through informal resolution will be settled through binding individual arbitration administered by the American Arbitration Association (AAA) in accordance with its Commercial Arbitration Rules. The arbitration will be conducted by a single arbitrator, and the arbitrator's decision will be final and binding.

Arbitration proceedings will be conducted in English. For claims under $10,000, the arbitration will be conducted based solely on written submissions unless the arbitrator determines a hearing is necessary. For larger claims, hearings will be conducted by video or telephone unless the arbitrator determines an in-person hearing is necessary, in which case it will be held in Wilmington, Delaware.

exceptions to arbitration

This arbitration provision does not apply to: (a) claims relating to intellectual property rights; (b) requests for injunctive or equitable relief; or (c) claims that qualify for small claims court in your jurisdiction, provided such court does not permit class actions.

class action waiver

You and inference.sh agree to waive any right to a jury trial and agree that disputes will be resolved solely on an individual basis and not as part of any class action, collective action, or representative proceeding. The arbitrator may not consolidate claims or preside over any form of class or representative proceeding.

mass arbitration procedures

If 25 or more claimants submit Dispute Notices raising similar claims with the same or coordinated counsel, these will be considered "Mass Arbitrations" and subject to the following procedures:

Bellwether Process: Counsel for both sides will each select 15 claims (30 total) to proceed first as test cases. Only these selected claims may be filed with AAA initially.
Mediation: After the bellwether arbitrations conclude, the parties will participate in a single mediation of all remaining claims, with the mediator's fee paid by inference.sh.
Remaining Claims: If mediation does not resolve all claims and 100 or more remain, either party may elect within 30 days to have remaining claims proceed in court rather than arbitration. If fewer than 100 claims remain, AAA will randomly select batches of 50 claims to proceed until all are resolved.

Statutes of limitations will be tolled for claims asserted in Mass Arbitrations until they are selected for arbitration or released to court.

severability

If any portion of this arbitration provision is found unenforceable, the remaining provisions will continue in effect. If the class action waiver is found unenforceable for a particular claim, that claim must be litigated in court while individual claims proceed in arbitration.

Nothing in this section limits mandatory consumer protections applicable in your jurisdiction.

model-specific terms

Certain AI models and Third-Party Services available through inference.sh may be subject to additional license terms, restrictions, or usage requirements imposed by their respective providers ("Model Terms"). By using any such model, you agree to comply with all applicable Model Terms in addition to these Terms.

Model Terms may include restrictions on:

Commercial use or specific industries
Use for training competing AI models
Generation of certain content types
Redistribution or modification of model outputs
Geographic availability or export controls

It is your responsibility to review and comply with the Model Terms applicable to any model you use. Model Terms are typically displayed on the model's page or documentation within the Services. If there is a conflict between these Terms and applicable Model Terms, the more restrictive provision applies.

inference.sh does not warrant that all models comply with all laws in all jurisdictions. You are solely responsible for ensuring your use of any model complies with applicable laws and Model Terms.

contact

Inference Shell Inc. Delaware, USA

For questions about these Terms: [email protected] For support: [email protected]