Terms of Service

Version 1.0. Effective date: 12 October 2026.

Previous versions of these terms remain available at weflayr.com/terms/archive.

Weflayr SAS ("Weflayr", "we", "us") operates the Weflayr platform, together with our website at weflayr.com, our APIs, dashboards, documentation and software development kits (the "Service"). These Terms of Service (the "Terms") are a binding contract between you ("Customer", "you") and Weflayr governing your access to and use of the Service.

By ticking the acceptance box at sign-up, or by accessing or using the Service, you agree to be bound by these Terms, our Data Processing Agreement and our Privacy Policy (together, the "Agreement"). If you do not agree, you must not use the Service.

Weflayr SAS is a société par actions simplifiée registered at 12 place Henri Bergson, 75008 Paris, France. RCS Paris 103 332 649. VAT FR24 103 332 649.

The Service is offered to businesses. It is not offered to consumers. By creating an account you confirm that you act for professional purposes.

1. The Service

The Service currently includes:

• Benchmarking. Comparison of candidate models against samples of your own prompts, evaluated by automated judges, to show which model preserves quality at lower cost for each of your use cases. Benchmark inference is run by us through the providers listed in our Data Processing Agreement and is paid from your credits. See section 6.
• Routing engine. The decision layer of the Gateway. You define a routing strategy: a preferred mix of models, the quality threshold that must be preserved against your preferred models, the business conditions under which the strategy applies (for example, customers whose AI margin is below a level you set) and any exclusions. For each request, the routing engine compares it with similar calls already benchmarked on your traffic and selects, among your approved models, the one that delivered the best balance of quality and cost on those calls, according to the trade-off you have chosen. Every routing decision is logged and can be audited. See section 7.
• AI Gateway. An API endpoint through which your application sends its production AI requests. The Gateway forwards each request to an approved model, applies the routing rules you configure, and returns the provider's response. You pay for inference either from your credits or through your own provider keys. See sections 7 and 8.
• Observability. Capture of metadata about the AI calls your application makes, as described in section 3, matched against the revenue you upload or connect from your billing provider, and reporting on cost, usage and profitability per customer, per feature and per model.

You connect your application to the Service through our SDKs or directly through the Gateway API.

We improve the Service continuously and may add, modify or discontinue features and models. Where we describe something as planned, in preview or in beta, it is not a commitment and is provided without warranty.

2. Eligibility and accounts

You must be at least 18 and use the Service lawfully. If you are agreeing on behalf of an organization, you confirm you have authority to bind it.

To use the Service you must:

• create an account with accurate and complete information
• keep your account credentials and API keys confidential
• tell us immediately at security@weflayr.com if you suspect unauthorised access
• accept responsibility for all activity under your account, including that of your users and of any application that uses your API keys

You are responsible for who you give access to, for the permissions you grant them, and for removing access when they no longer need it.

One organization, one account. You may hold only one account per legal entity, with as many users and projects as you need within it. Creating multiple accounts to circumvent usage limits or to obtain promotional credits more than once is prohibited and may lead to the suspension of all accounts concerned. We may ask you to verify your identity, your organization or your payment method before granting access to certain features or to higher limits.

3. Your data

"Customer Data" means the data submitted to or generated by the Service under your account. It comprises:

• Call metadata, being the identifier you assign to your customer or user, the feature name or tag you assign, the model and provider used, consumption metadata such as input, output, reasoning and cached token counts, the cost computed from that consumption, latency, the routing decision taken by the Gateway, and a timestamp.
• Revenue data, being the amounts you attribute to your customers, whether uploaded by you or read from your billing provider.
• Prompt and completion content, as described in section 5, whether captured by the SDK or transmitted through the Gateway.
• Provider credentials, being the API keys of third-party model providers that you register with us under section 8.
• Account, user and configuration data, including your routing rules, your approved models and your logging settings.

The authoritative list of fields, and how each is treated, is set out in Annex A of our Data Processing Agreement.

You choose the identifier. The Service does not require a name, an email address or any other direct identifier. You decide what value is passed in the customer identifier field, and we recommend an opaque or pseudonymous value. You are responsible for that choice and for the lawful basis for any personal data it contains.

Connecting your billing provider. You may provide revenue data by uploading it, or by connecting your billing provider. We currently support Stripe. To connect it, you create a restricted API key in your own Stripe account and configure it in the Service. Records read from your billing provider may contain personal data relating to your own customers, including names and email addresses. You are responsible for the lawful basis for that processing and for informing the individuals concerned where required.

You keep all rights in Customer Data. You grant us a non-exclusive, worldwide, royalty-free licence to host, store, process and transmit Customer Data solely to provide, secure and maintain the Service, to comply with a legal obligation, and as permitted by the Data Processing Agreement. This licence ends when the Agreement ends, subject to the deletion periods in section 10.

You confirm that you have the rights and the lawful basis necessary to send us Customer Data and to authorise the processing described in the Agreement.

You must not send us payment card data, government identifiers, children's data, special category data within the meaning of Article 9 of the GDPR, or criminal conviction data within the meaning of Article 10, whether through the SDK or through the Gateway.

4. What we do with your data

We use operational metadata, being consumption, cost, latency, model, routing and outcome data, to operate, secure and improve the Service, including to improve our routing and evaluation models. We may also produce aggregated statistics from usage across our customers, for example benchmarks of cost per feature by category of application or model performance rankings, to improve the Service and to publish market insights. Such statistics never identify you, your customers or any individual, and are never presented in a way that could.

We do not use the content of your prompts and completions to train, fine-tune or evaluate any machine learning model, whether ours or a third party's. We never sell, rent or licence Customer Data.

We disclose Customer Data only to the providers we use to operate the Service, which are listed in Annex B of our Data Processing Agreement and, for model providers, at weflayr.com/models, and where the law requires it.

Prompt and completion content is disclosed to model providers only in two cases: to the model you have approved, when the Gateway forwards your request to it under section 7, and to the candidate models we test for you under section 6.

5. Prompt and completion content

The Service always processes metadata about your AI calls. Whether it also keeps the text of your prompts and of the completions returned to you depends on your settings.

Through the SDK. Capture of prompt and completion content is on when you install the SDK, so that benchmarking operates on your real traffic from day one rather than on samples you construct. Setting capture_message_content to false in your SDK configuration strips that content before anything is exported from your servers, ensuring it never reaches us. In that configuration, we receive metadata only.

Through the Gateway. The Gateway must receive the full content of each request in order to forward it to the model provider and return the response. That transit is inherent to the Service and cannot be disabled. Whether we also log the prompt and output content is your choice. Content logging is on by default and can be turned off per API key in your dashboard. When logging is off, request and response content is held in memory only for the time needed to serve the request and is not written to storage. Metadata is always retained in every configuration, because we need it to bill, to route and to secure the Service.

You decide what your application sends and what settings you keep. You are responsible for the lawful basis for that content, for informing the individuals concerned where required, and for assessing whether a data protection impact assessment is needed.

6. Benchmarking and candidate models

Benchmarking compares candidate models against samples of your own prompts. It operates on the content captured or logged under section 5.

What we send. We send samples of your prompts to candidate model providers, receive their outputs, and evaluate those outputs with automated judges to compare quality and cost.

Which providers. Benchmark inference is run by us, on our own provider accounts. The candidate models, their provider, their hosting region and the provider's published position on retention and training are listed at weflayr.com/models. You may restrict benchmarking to a subset of those models at any time.

What it costs. Benchmark inference is charged to your credits at the provider's list price for the tokens consumed, plus the benchmark fee published on our pricing page. Your own provider keys are not used for benchmarking. You see the estimated cost of a benchmark before you launch it.

Benchmark results are estimates based on samples. They do not guarantee that a change of model will reduce your cost or preserve your quality in production. You remain responsible for deciding which models to approve.

7. The AI Gateway

The Gateway receives the AI requests your application sends to our endpoint, selects a model according to your routing rules, forwards the request to that model's provider, and returns the response to you.

Approved models. You configure the list of models your organization permits. The Gateway routes only within that list, including on failover, load balancing and automatic routing. It never substitutes a model, a provider or a hosting region outside your approved list. If no approved model is available, the request fails and we tell you why.

Routing rules. You define your routing strategy: your preferred models, the quality threshold measured in your benchmarks, the cost or margin conditions, and any exclusions. The Gateway applies your rules to each request using the business context your application passes. Routing decisions are logged as call metadata so you can audit them.

Hosting region. For every model we publish a hosting region in the dashboard and in the model metadata returned by the API. Where that region is the European Union, inference for that model is performed there and we have confirmed it with the provider. Where the region is marked global, the provider may serve the request from any region in its own infrastructure.

How we reach providers. We access models through cloud platforms in European regions or direct agreements with the model providers. Each channel is listed at weflayr.com/models and is subject to section 11..

Model availability. Models are made available by their providers and may be changed, degraded or withdrawn by them at any time. We do not guarantee the availability, latency or output of any model. We may add or remove models from the catalogue. Where we remove a model from your approved list we will tell you before doing so, except where the provider withdraws it without notice to us.

Inference prices. Inference consumed through the Gateway from your credits is charged at the provider's list price as published in the Service at the time of the request. We may change the list price of a model when its provider does, with effect from the date of the provider's change. Other price changes follow section 15.

8. Bring your own keys

You may register your own API keys for third-party model providers ("Provider Credentials") so that the Gateway calls those providers on your account rather than on ours.

How we hold them. Provider Credentials are encrypted at rest and stored in a managed secret store. They are used solely to make the requests your application sends through the Gateway, and are never used for benchmarking, for other customers or for any purpose of our own. You may rotate or delete them at any time from your dashboard. Deleting a credential takes effect for subsequent requests.

Your relationship with the provider. When you use your own keys, the inference contract is between you and the provider. The provider acts as your own processor or service provider, under your agreement with it, and not as our sub-processor. You are responsible for complying with the provider's terms, for the spend incurred on your provider account, for any rate limits the provider applies to your keys, and for the provider's handling of your data.

Our role. We transmit your requests and responses, apply your routing rules, record call metadata and, where logging is on, content, and debit the platform fee from your credits under section 15. We are not liable for a provider's suspension of your keys, for its billing, or for its availability.

9. Availability and support

We monitor the Service continuously and work to keep it available. We do not guarantee uninterrupted availability unless a service level agreement is agreed in writing with you. The availability of any model depends on its provider and is outside our control.

We may carry out maintenance and will give notice where it is likely to affect you. Our current status is published at status.weflayr.com.

Support is available at support@weflayr.com. We aim to respond within one business day.

10. Where we store your data, and for how long

All Customer Data stored by us is stored in the European Union. The table below matches Annex A of our Data Processing Agreement.

Encrypted backups are retained for 30 days and replicated within the European Union. Where content is stored, it can persist in backups for up to 30 days after the retention period above has expired, and for up to 30 days after deletion on closure.

Where content logging is off we hold no prompt or completion content for the keys concerned and are therefore unable to search for, extract, rectify or delete such content.

11. Sub-processors

The infrastructure providers we use to operate the Service are listed in Annex B of our Data Processing Agreement, with their location and role. Model providers and their models are listed in our model catalogue at weflayr.com/models (the "Model Catalogue")

Adding or replacing a provider. We will give you at least thirty days notice before we engage a new provider or replace an existing one. If you reasonably object on data protection grounds, you may tell us within that period, and we will either propose an alternative or you may terminate the affected part of the Service without penalty.

AI providers and models. AI providers and their models are chosen by you and are not subject to the notice above. We may add AI providers and models to our catalogue, or remove them, at any time. A new AI provider or model is never added automatically to your approved list or to your benchmark list, and none of your Customer Data reaches it until you, or a user you have authorised, select it. Selecting a model is your specific authorisation for its provider to process your Customer Data under the conditions published for that model. Your approved list is the control that restricts which providers your production traffic reaches; your benchmark list is the control for candidate models. You can remove a model from either list at any time, with effect for subsequent requests.

Changes to a model you have selected. If the hosting region of a model on your approved list or your benchmark list changes, or if its provider changes its published position on retention or training, we will tell you by email or in the Service without undue delay after becoming aware of it, so that you can decide whether to keep it.

12. Security

We maintain an information security management system aligned with ISO/IEC 27001 and SOC 2 Type II and implement the technical and organizational measures set out in Annex C of our Data Processing Agreement.

This includes encryption in transit and at rest, role-based least-privilege access with multi-factor authentication, logical isolation of customer data, storage of Provider Credentials and of any credentials you provide for a connected integration in a managed secret store, continuous control and threat monitoring, dependency scanning on every change, and an annual application penetration test by an independent third party.

Our current security posture, our certification and audit status, and the position of each sub-processor are published at trust.weflayr.com. We make no representation of certification beyond what is stated there.

No system is completely secure, and we cannot guarantee absolute security.

13. Security incidents

If we become aware of a personal data breach or of a security incident affecting your data, including a compromise of your Provider Credentials, we will notify you without undue delay and in any event within forty-eight hours of becoming aware of it.

We notify your registered account email address. Our notification will describe the nature of the incident, the categories of data affected, the likely consequences and the measures taken or proposed, to the extent then known, and will be supplemented as the investigation progresses. We will cooperate with you and with the relevant authorities and provide reasonable assistance with your own notification obligations.

14. Data protection

Our Data Processing Agreement is incorporated by reference into and forms part of these Terms. It applies to every customer in the version current at the time of your use, and it does not require separate signature to be binding. We do not negotiate amendments to it other than as part of a master services agreement signed by us. Where the Data Processing Agreement and these Terms conflict on a data protection matter, the Data Processing Agreement prevails.

Where you have signed a separate data processing agreement with us, that agreement applies in place of our standard Data Processing Agreement for as long as it remains in force.

Our Privacy Policy explains how we handle personal data for which we are the controller, including the data of your administrators, of your billing contacts and of visitors to our website.

15. Credits, fees and payment

Prepaid credits. Paid features of the Service are consumed from a prepaid balance of credits attached to your organization ("Credits"). Credits are denominated in United States dollars. One Credit equals one US dollar of value in the Service. Credits are a prepayment for the Service, usable only within the Service. They are not a deposit, bear no interest, are not transferable to another organization, and cannot be redeemed for cash.

Topping up. You buy Credits by card through our payment processor, Stripe, from your dashboard. We may set a minimum and a maximum top-up amount, published on our pricing page. You see the amount you will be charged and the Credits you will receive before you confirm. Other payment methods may be offered later; where we do, we will say so on the pricing page.

Paying in your own currency. You may pay in the currency of your card, where offered. The amount in your currency is calculated by Stripe at the exchange rate shown to you before you confirm, and your balance is credited in US dollars. The conversion is final once the payment is made, and your balance is not adjusted for later movements in exchange rates. Your card issuer may apply its own charges, which are outside our control.

Platform fee. Our remuneration is a platform fee, at the rates published on our pricing page at the time of the transaction:

• where inference is paid from your Credits, the fee is charged on each top-up and shown before you confirm, so that the inference you consume is debited at the provider's list price;
• where you use your own Provider Credentials, the fee is calculated on the list price of each request routed through the Gateway and debited from your Credits;
• benchmark inference is debited at the provider's list price plus the benchmark fee.

Auto-recharge. You may enable automatic top-ups so that your balance is replenished when it falls below a threshold you choose. When enabled, you authorise us to charge your saved card for the recharge amount you have set each time the threshold is crossed. You can change the threshold and the amount, or disable auto-recharge, at any time from your dashboard. We send you a receipt for each automatic top-up and we tell you by email if a charge fails. We are not liable for interruptions of the Service caused by a failed automatic top-up.

Consumption. Credits are debited as you use the Service, at the rates and list prices shown in the Service at the time of the request. Your balance and your consumption history are available in your dashboard at all times. If your balance reaches zero, requests paid from Credits are rejected until you top up; requests using your own Provider Credentials continue only as long as your balance covers the platform fee. We may let your balance go slightly negative to absorb in-flight requests; any negative balance is deducted from your next top-up.

No refunds. Credits are purchased for your business use and are not refundable, in whole or in part, except where we terminate the Agreement for our convenience under section 26 or where a refund is required by law. If you believe a charge or a debit is incorrect, tell us at support@weflayr.com within thirty days of the transaction and we will investigate and correct any error.

Expiry. Credits expire after twelve months of inactivity. Expired Credits are forfeited without refund. We remind you by email before Credits expire.

Promotional credits. We may grant free or promotional Credits at our discretion. They are limited to one grant per organization, may expire on the date we state, cannot be transferred or redeemed for cash, are consumed before paid Credits, and may be withdrawn if we reasonably believe they were obtained through multiple accounts, false information or other abuse.

Failed payments and chargebacks. If a card payment fails or is reversed after Credits have been consumed, the corresponding amount becomes immediately payable. We may suspend your access until it is paid and we may recover it from your next top-up. Initiating a chargeback for a payment you authorised is a material breach of the Agreement.

Taxes. Prices and fees are exclusive of VAT and of any other applicable tax. VAT is applied according to your country and your VAT status as declared in your account. Where you are a business established outside France and provide a valid VAT identification number, no French VAT is charged and the reverse charge mechanism applies. You are responsible for the accuracy of the tax information you provide.

Invoices and receipts. We issue an invoice for each top-up and provide a monthly consumption statement in your dashboard.

Price changes. We may change the platform fee with at least fourteen days notice by email or in the Service. Credits already purchased remain usable at the new rates. Provider list prices follow section 7.

Agreements in writing. Where we have agreed a subscription, a minimum commitment, a pilot or an enterprise plan with you in writing, that agreement governs fees and payment terms in place of the provisions above, and section 15 applies to anything it does not cover. Invoices payable in arrears under such an agreement are due within thirty days of issue. In accordance with article L441-10 of the French Commercial Code, any sum unpaid on its due date bears interest at the European Central Bank refinancing rate plus ten percentage points, applied automatically and without prior notice, together with a fixed indemnity for recovery costs of forty euros per invoice. Where recovery costs exceed that amount, we may claim the excess on production of supporting evidence. If an invoice remains unpaid after we have notified you, we may suspend the Service after reasonable notice.

16. Usage limits

We may apply technical limits, including rate limits per API key and concurrency limits, to protect the Service and other customers. We will tell you before applying a limit that materially affects you, except where immediate action is needed. Model providers apply their own limits, which we pass on.

You must not circumvent usage limits, including by creating multiple accounts or by splitting traffic across organizations.

17. Acceptable use

You must not use the Service to break the law, to infringe anyone's rights, to send data you have no right to send, to attempt to access another customer's data, to probe, scan or disrupt the Service or its security measures without our permission, to circumvent usage limits, or to resell access without our written consent.

You must not use the Gateway to carry out prompt injection, jailbreaking, red teaming or any other adversarial action against a model in breach of its provider's terms.

You must not use the Service, or the data and benchmarks it produces, to develop a competing product, or to train a model that replicates our routing or evaluation functionality.

You are responsible for the acts and omissions of your users and of the applications that use your API keys. We may suspend an account that puts the Service, a model provider relationship or other customers at risk, and we will tell you why.

18. Model terms

Models are provided by third-party model providers under their own terms, usage policies and documentation ("Model Terms"). By approving a model, or by sending a request to it through the Gateway, you agree, and you will ensure that your users agree, to comply with the Model Terms applicable to that model.

You are responsible for reviewing the Model Terms before approving a model and for determining whether they allow your intended use, including any regulated, high-risk or customer-facing use. Some providers restrict access from certain countries or for certain organizations. You must not use the Service to reach a model you are not permitted to access, including through a proxy or a virtual private network.

We may suspend or restrict your access to a model if we reasonably believe that you or your users have breached its Model Terms, or if its provider requires us to. We will tell you in advance where we can. Each provider retains sole control over access to its models, and we are not liable for a provider's suspension, restriction, withdrawal or modification of a model.

As between you and us, these Terms govern. Model Terms govern your use of the model to the extent the provider requires. We do not modify, waive or limit any Model Terms.

19. Sanctions and export control

You must not use the Service in or from, or make it available to anyone in, a country or territory subject to comprehensive sanctions of the European Union, the United Kingdom or the United States, or to any person listed on the sanctions lists of those jurisdictions. You confirm that you are not such a person and are not owned or controlled by one. You will comply with applicable export control laws, including the European Union dual-use regulation, in your use of the Service and of the models it gives access to. We may suspend or terminate access where we reasonably believe this section is breached.

20. Our intellectual property

The Service, including its software, routing and evaluation models, benchmark methodology, design and documentation, is owned by Weflayr or its licensors. You may use it while your account is open and in good standing, and not otherwise. We reserve all rights not expressly granted.

You must not copy the Service, reverse engineer or attempt to derive the source code or logic of our routing, benchmarking, evaluation or optimization models, or systematically extract our benchmark data, model rankings or cost intelligence.

Our SDKs are open source and licensed separately under the Elastic License 2.0, which has only three high-level limitations. You cannot:

• provide the products to others as a managed service
• circumvent the license key functionality, or remove or obscure features protected by license keys
• remove or obscure any licensing, copyright or other notices

Outputs. As between you and us, you own the outputs returned to you by model providers through the Gateway, subject to the applicable Model Terms. We claim no rights in them.

21. Feedback

If you give us suggestions or feedback about the Service, you grant us a perpetual, irrevocable, royalty-free licence to use and incorporate it without restriction or obligation to you.

22. Confidentiality

"Confidential Information" means non-public information disclosed by one of us to the other that is identified as confidential or that a reasonable person would understand to be confidential. Our Confidential Information includes the non-public aspects of the Service and our product plans. Your Confidential Information includes your Provider Credentials, your routing rules and your benchmark results.

Each of us will protect the other's Confidential Information with at least reasonable care, use it only in connection with the Agreement, and limit access to those who need it and are bound by equivalent obligations. This does not apply to information that is or becomes public without breach, was already lawfully known, is lawfully received from a third party, or is independently developed. Either of us may disclose where compelled by law, giving prior notice where legally permitted.

23. Warranties and disclaimers

We provide the Service with reasonable skill and care.

Beyond that, and to the fullest extent permitted by law, the Service is provided as is and as available, without other warranties, express or implied.

We do not warrant that the Service will be uninterrupted or error-free, that any benchmark, routing decision, recommendation or estimate will be accurate or will produce a saving, or that any output obtained from a model provider will be accurate, complete, lawful or fit for any purpose. Outputs are generated by third-party models and may be wrong. You are solely responsible for evaluating outputs, for implementing human review and safeguards appropriate to your use case, for your routing rules and approved models, and for deciding what to change in your production systems.

We make no representation about any model provider's data handling, retention, training, security, availability or intellectual property practices beyond reproducing the provider's own published position.

Nothing in the Agreement excludes or limits liability that cannot be excluded or limited by law, including for fraud, for wilful misconduct, or for death or personal injury caused by negligence.

24. Limitation of liability

Neither of us is liable for indirect or consequential loss, or for loss of profits, revenue, data, goodwill or business opportunity, however caused.

Our total aggregate liability arising from or in connection with the Agreement, whether in contract, tort or otherwise, shall not exceed the greater of five hundred euros and the platform fees and benchmark fees you paid us in the twelve months preceding the event giving rise to the claim. Amounts paid for Credits that were passed on to model providers as inference cost are not fees paid to us for this purpose.

The limits in this section are aggregate limits covering all claims under these Terms and under any data processing agreement between us, whether our standard Data Processing Agreement or one signed separately, taken together. Neither a data processing agreement nor any separately agreed pricing creates a separate, additional or increased cap unless expressly agreed in a master services agreement signed by us.

These limitations reflect the allocation of risk between us and form an essential part of the bargain.

25. Indemnity

You will defend us against claims arising from your use of the Service in breach of the Agreement or of Model Terms, from Customer Data that infringes third-party rights or breaches applicable law, from the use you make of outputs, and from your users' acts and omissions.

We will defend you against claims that the Service, excluding third-party models and their outputs, infringes a third party's intellectual property rights, provided you tell us promptly, let us control the defence, and give us reasonable cooperation. This does not cover claims arising from Customer Data, from a model provider's output or from your combination of the Service with anything we did not supply.

26. Term, termination and deletion

The Agreement runs from your acceptance until terminated in accordance with this section.

You may close your account at any time from the dashboard or by emailing support@weflayr.com. Either of us may terminate for material breach that is not remedied within thirty days of written notice, or immediately where the breach cannot be remedied. We may suspend or terminate where required by law, or where a model provider requires us to stop serving you.

Unused Credits. When you close your account, or when we terminate for your breach, unused Credits are cancelled without refund. If we terminate the Agreement for our convenience, or discontinue the Service, we will refund the unused paid Credits on your balance, excluding promotional Credits, within thirty days. Where we have agreed a committed term with you in writing, fees for that term are payable in full and are not refunded or prorated, except where you terminate for our material breach.

On closure or termination, your access ends and we delete Customer Data within thirty days, with encrypted backups rolling off within a further thirty days. If you ask us before that period expires, we will export your data in a commonly used machine-readable format, and we will confirm deletion in writing on request. We retain data beyond these periods only where the law requires it, in particular the accounting records referred to in section 10.

27. Changes to the Service and to these Terms

We may update the Service or these Terms. Where a change materially affects you, we will give you at least thirty days notice by email or in the product before it takes effect, and we will state at the top of these Terms what has changed. Continuing to use the Service after the effective date means you accept the change. If you object, you may close your account under section 26; Credits remain usable until the effective date and are then cancelled without refund.

Changes to the model catalogue, to provider list prices and to default usage limits are operational changes published in the Service and are not subject to the notice above.

28. General

The Agreement, comprising these Terms, the Data Processing Agreement, the Privacy Policy, the pricing page and any terms separately agreed in writing between us, is the whole agreement between us and supersedes all prior agreements on its subject matter. Where you have signed a master services agreement with us, that agreement prevails over these Terms in the event of a conflict.

You may not assign the Agreement without our prior written consent. We may assign it to an affiliate or in connection with a merger, acquisition or sale of assets, and we will tell you if we do.

If a provision is held unenforceable, the rest continues to apply. A failure to enforce a provision is not a waiver of it.

Neither of us is liable for delay or failure caused by events beyond reasonable control, including the unavailability of a model provider.

29. Governing law

The Agreement, and any dispute or claim including non-contractual disputes, is governed by French law. The competent courts of Paris, France have exclusive jurisdiction.

30. Contact

Weflayr SAS, 12 place Henri Bergson, 75008 Paris, France

General and contractual questions: contact@weflayr.com

Security: security@weflayr.com

Support: support@weflayr.com

Data protection, including Data Processing Agreement requests: privacy@weflayr.com

Our current security and certification status is published at trust.weflayr.com.