Last updated: August 5, 2026
By creating an account or using InfoPlatform.ai ("IP.ai", formerly ModelForge), you agree to these Terms of Service. If you do not agree, do not use the service.
InfoPlatform.ai is a platform that enables businesses to fine-tune open-weight AI models (such as Kimi K2.6, DeepSeek V3.1, Qwen 3.5, Nemotron 3, GPT-OSS, and Inkling) on their own data, deploy them via an OpenAI-compatible API compatible with tools like OpenCode and Cursor, and improve them with feedback loops. The service includes data upload, model selection, training, inference API access, and a feedback collection system.
In addition to the hosted platform, InfoPlatform.ai offers an optional professional-services engagement, available on request, to implement the training and self-served inference stack inside a customer's own infrastructure ("in-house implementation"). In-house engagements are governed by a separate statement of work; these Terms apply to the hosted platform.
You must be at least 18 years old to use InfoPlatform.ai. You are responsible for maintaining the security of your account credentials. You must not share your API keys.
No Free Tier or Trial: An account can register, upload data, and explore, but training, inference, and API access require an active paid plan or a custom contract. There is no free tier and no free trial.
Paid Plans: Plans are billed monthly via Stripe as a base platform fee that includes a monthly allowance of determinations. Training is included in all paid plans. You can upgrade, downgrade, or cancel at any time.
What Counts as a Determination: A determination is one completed answer delivered to you: the model produced a response to your request. Where a single request causes several internal model calls, for example an agent that consults a tool and reasons in more than one step, that still counts as one determination. We do not charge you for a request that returns no answer, for a request we refuse, or for a request that stops at a human review step instead of producing an answer. Your current count and any overage are shown on the Billing page in your account.
Overage: Determinations past your monthly allowance are billed at the per-determination rate published for your plan. Going past the allowance does not interrupt your service. Separate technical limits on total tokens, training runs, dedicated GPU runs, and stored Knowledge Base chunks exist to prevent runaway usage; these are documented on the Billing page and are not additional charges.
Custom Contracts: Higher-volume, in-house, and tailored engagements are available by contract, arranged on a call and governed by a separate agreement (and, for in-house deployments, a statement of work).
Refunds: If you are unsatisfied within the first 14 days of a paid subscription, contact us for a full refund.
Cancellation: Cancel via Billing > Manage Billing. Your models remain accessible until the end of the billing period. You keep the weights of any open-weight model you fine-tuned forever.
Your Data: You retain full ownership of all data you upload. We use it only to provide the service to you, and never to train our own models or any other customer's. Providing the service involves specific, named sub-processors for training, cleaning, document transcription and search indexing: section 3 of the Privacy Policy lists exactly which features send what, and to whom, and section 3a sets out the configuration that keeps data within infrastructure we operate. Read it before uploading anything restricted.
Open-weight models: You own the fine-tuned adapter weights of any open-weight model you train. Depending on the active training backend they are stored in InfoPlatform.ai-controlled object storage or as artifacts in InfoPlatform.ai's Weights & Biases account; in either case you may request an export at any time and keep full ownership even if you cancel your subscription (export is fulfilled by us on request; there is no self-serve download endpoint yet), subject to the base model's underlying license (e.g. MIT, Apache 2.0, or the Llama Community License).
Data Privacy: All data is encrypted at rest (AES-256) and in transit (TLS). MCP server credentials are encrypted with AES-256-GCM.
You agree not to:
We reserve the right to suspend accounts that violate these terms.
Models trained on InfoPlatform.ai are AI systems. Their outputs may be inaccurate, incomplete, or biased. You are responsible for reviewing AI-generated content before use. InfoPlatform.ai does not guarantee the accuracy, reliability, or fitness for purpose of any trained model.
Some customers use InfoPlatform.ai for regulated classification work: export control numbers, CUI identification under CMMC, tariff codes, restricted-party screening. These terms override any marketing statement to the contrary.
If your obligations require that data never leaves your own network, ask us about an in-house implementation before uploading anything.
An agent is a workflow you define that runs your model and, where you configure it to, calls tools and external services through Connections. You choose which services an agent may reach and supply the credentials for them. Those credentials are encrypted and scoped to your account.
You are responsible for what your agents do. An agent acts under your authority and within the permissions of the credentials you provide. Before connecting a service, confirm that automated access is permitted under your agreement with that provider and that the credential carries no more privilege than the agent needs.
Human approval gates are yours to place. You may put a human review step anywhere in an agent. When a run reaches one it pauses and waits in your approval inbox until a person decides; it is not charged as a determination while it waits. Whether a given workflow requires human review, and who is competent to give it, is your decision. We do not review your agents, and an agent that runs without a human gate is running exactly as you configured it.
If you are a forward deployed engineer, consultancy, or contractor using InfoPlatform.ai to build for someone else, the organisation whose data is uploaded is the data owner, and the account holding that data is theirs. Access for you is granted by them and can be withdrawn by them at any time, immediately and without our involvement.
Each client is a separate tenant. Data, models, weights, knowledge bases, and access trails are scoped to one client account and are never pooled across the clients you work with. Every action you take on a client account is recorded against you in that client's access trail, which is theirs to read.
Withdrawing your access does not remove their work. The client keeps their data, their trained weights, and their endpoint whether or not you continue to have access. If you need us to transfer an engagement to the client's own administration, contact us and we will arrange it.
You are responsible for your authority to act. Before uploading a client's material you must be permitted to do so under your agreement with them. We cannot verify the terms of your engagement.
TO THE MAXIMUM EXTENT PERMITTED BY LAW, MODELFORGE'S AGGREGATE LIABILITY SHALL NOT EXCEED THE TOTAL FEES PAID BY YOU IN THE 12 MONTHS PRECEDING THE CLAIM. MODELFORGE SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES.
We may suspend or terminate your account if you violate these terms. You may request account deletion at any time via Settings. There is a 30-day grace period during which you can cancel the deletion request. After 30 days, all data is permanently deleted (except billing records retained for legal requirements).
These terms are governed by the laws of the State of Delaware, United States, without regard to conflict of laws principles.
We may update these terms. We'll notify you of material changes via email at least 30 days before they take effect. Continued use after changes constitutes acceptance.
Questions about these terms: support@infoplatform.ai