The contractual framework that governs your use of OwnLLM: the SaaS control plane, the native desktop agent, and the network relay.
Last updated: May 4, 2026
These Terms of Service form a binding agreement between you, acting on behalf of your organization (the "Customer"), and OwnLLM, operated by Impulse Lab, a French entity ("OwnLLM", "we", "us"). By creating an account, paying a subscription, installing the OwnLLM desktop application, or otherwise using the service, you confirm that you have authority to bind the Customer and that the Customer accepts these terms. If you do not agree, do not use the service.
{slug}.ownllm.app.OwnLLM provides software that turns a Customer-operated GPU machine into a private multi-user AI platform. The Service includes the SaaS dashboard, single sign-on, API key management, audit logging, an OpenAI-compatible proxy, the native agent, and a Cloudflare-backed network relay that lets the Customer expose the agent to its users without opening inbound ports.
Inference runs entirely on the Customer machine. OwnLLM does not host model weights and does not run inference on its infrastructure.
The Customer designates an administrator at signup. The administrator may invite Users, assign roles, configure SSO, generate pairing keys, and revoke access. Each User must authenticate with credentials issued specifically to them; shared accounts are not permitted. The Customer is responsible for keeping credentials, pairing keys, and API keys confidential and for all activity under its account.
Subscriptions are sold as monthly or annual plans (Team, Startup, Enterprise) at the prices shown on ownllm.app/#pricing. Fees are charged in advance through Stripe. Annual plans receive a documented discount and are non-refundable except as required by law. Monthly plans renew automatically until cancelled and may be cancelled at any time, taking effect at the end of the current billing cycle.
Prices are quoted excluding VAT and any other applicable taxes, which are added based on the Customer's billing address. Late payments incur the legal interest rate plus a fixed €40 recovery indemnity under French commercial law. The Customer is responsible for the cost of the GPU machine, its electricity, and its internet bandwidth.
We may offer free trials, beta features, or early access programs. Beta features are clearly labeled and are provided "as is" without any warranty or service-level commitment. We may modify, suspend, or remove beta features at any time.
The Customer is responsible for:
The Customer and its Users may not:
We may suspend access without prior notice in case of security risk, ongoing abuse, or unpaid invoices, with a summary explanation by email.
OwnLLM curates a catalog of open-weight models accessible through the agent (Mistral, Llama, Qwen, DeepSeek and others). Each model is distributed under its own license, and that license governs the Customer's right to use, copy, fine-tune, or distribute the model. The Customer is responsible for reviewing and complying with each model's license, including any commercial-use thresholds, branding requirements, or prohibited-use clauses.
Models are provided by their respective publishers without warranty by OwnLLM. We do not warrant that any model will produce accurate, complete, current, or non-infringing output.
OwnLLM and its licensors retain all intellectual property rights in the Service, including the source code of the control plane, the desktop agent, the documentation, and the trademarks. Subject to these terms, OwnLLM grants the Customer a non-exclusive, non-transferable, worldwide right to use the Service during the subscription term.
The Customer retains all rights in Customer Data and grants OwnLLM the limited rights needed to host, transmit, back up, and otherwise operate the Service. We do not use Customer Data to train or fine-tune any model.
Output generated by the models is deemed Customer Data, subject to the underlying model license.
Our handling of personal data is described in the Privacy Policy and in our DPA. We process personal data only on the Customer's documented instructions and only to deliver the Service. On request, we provide a data export and delete tenant data within 30 days of contract termination, except where retention is legally required.
Each party agrees to keep the other party's confidential information confidential, to use it only to perform the contract, and to protect it with at least the same care it uses to protect its own confidential information. The obligation survives the termination of the contract for three years.
We target 99.9% monthly availability for the SaaS control plane, excluding scheduled maintenance and force majeure. Tenant availability also depends on the Customer's GPU machine and internet connection, which OwnLLM does not control; we therefore do not commit to a tenant-wide availability SLA on the Team and Startup plans. A bespoke SLA may be negotiated under an Enterprise contract.
We warrant that we provide the Service with the level of care and skill reasonably expected of a professional operator. Except for this warranty and any warranty that cannot be excluded by law, the Service is provided "as is" and "as available". OwnLLM expressly disclaims warranties of merchantability, fitness for a particular purpose, non-infringement, and uninterrupted or error-free operation. Model outputs may be inaccurate or biased; the Customer remains solely responsible for evaluating their fitness for any given use.
To the maximum extent permitted by applicable law, neither party is liable for indirect, incidental, special, consequential, or punitive damages, including lost profits, lost revenue, lost data, or business interruption. Each party's aggregate liability arising out of or related to these terms is limited to the fees paid or payable by the Customer to OwnLLM during the twelve months preceding the event giving rise to the claim. These limitations do not apply to liability that cannot be excluded by law, including for fraud, gross negligence, willful misconduct, or breach of confidentiality or intellectual-property obligations.
The Customer will defend, indemnify, and hold OwnLLM harmless from any third-party claim arising from (a) Customer Data, (b) the Customer's violation of these terms or of any law, or (c) the Customer's use of any open-weight model in breach of the model's license.
Subscriptions remain in force until cancelled. The Customer may cancel at any time through the Stripe customer portal; the cancellation takes effect at the end of the current billing cycle. Either party may terminate immediately on written notice if the other party materially breaches the contract and fails to cure the breach within 30 days, or immediately if the other party becomes insolvent. Upon termination, the Customer's access to the Service ends, and the data-deletion provisions of section 11 apply.
We may update these terms to reflect product, legal, or regulatory changes. Material changes are notified at least 30 days before they take effect. Continued use of the Service after the effective date constitutes acceptance. Customers that do not accept material changes may terminate without penalty during the notice period and obtain a pro-rata refund of any prepaid fees.
These terms are governed by French law. Any dispute arising out of or in connection with these terms is submitted to the exclusive jurisdiction of the competent courts of Paris, France, except where mandatory consumer protection or data protection law requires otherwise.
Legal and contractual inquiries: legal@ownllm.app. General contact: /contact.