{"slug":"choose-a-private-model","id":"choose-a-private-model","type":"guide","title":"Choose a Private Model","description":"A decision playbook for choosing models and deployment patterns when privacy, compliance, sensitive documents, or internal data controls matter.","last_updated":"2026-04-24","last_verified":null,"verification_status":"unverified","markdown_url":"/content/guides/choose-a-private-model.md","html_url":"/guides/choose-a-private-model","api_url":"/api/v1/guides/choose-a-private-model.json","content_hash":"703bcb7f9a64d20ef79e1a185176b4b1cae366528ceb82ee6de1b6c98e378810","sha256":"703bcb7f9a64d20ef79e1a185176b4b1cae366528ceb82ee6de1b6c98e378810","tags":["privacy","models","compliance","local","playbook"],"relationships":{"links":[],"related":[{"id":"choose-a-local-model","title":"Choose a Local Model","type":"guide","html_url":"/guides/choose-a-local-model","markdown_url":"/content/guides/choose-a-local-model.md","shared_tags":["local","models","privacy","playbook"],"score":6},{"id":"best-for-task-matrix","title":"Best-For Task Matrix","type":"guide","html_url":"/guides/best-for-task-matrix","markdown_url":"/content/guides/best-for-task-matrix.md","shared_tags":["models","playbook"],"score":4},{"id":"choose-a-cheap-model","title":"Choose a Cheap Model","type":"guide","html_url":"/guides/choose-a-cheap-model","markdown_url":"/content/guides/choose-a-cheap-model.md","shared_tags":["models","playbook"],"score":4},{"id":"choose-a-research-model","title":"Choose a Research Model","type":"guide","html_url":"/guides/choose-a-research-model","markdown_url":"/content/guides/choose-a-research-model.md","shared_tags":["models","playbook"],"score":4},{"id":"choose-a-coding-model","title":"Choose the Best AI Coding Model","type":"guide","html_url":"/guides/choose-a-coding-model","markdown_url":"/content/guides/choose-a-coding-model.md","shared_tags":["models","playbook"],"score":4},{"id":"local-models-private-documents","title":"Use Local Models for Private Documents","type":"guide","html_url":"/guides/local-models-for-private-documents","markdown_url":"/content/guides/local-models-for-private-documents.md","shared_tags":["local","privacy"],"score":4}],"explicit":{}},"metadata":{"title":"Choose a Private Model","type":"guide","id":"choose-a-private-model","description":"A decision playbook for choosing models and deployment patterns when privacy, compliance, sensitive documents, or internal data controls matter.","last_updated":"2026-04-24","tags":["privacy","models","compliance","local","playbook"]},"content_text":"# Choose a Private Model\n\nPrivacy is a deployment requirement, not just a model property. The right answer may be local hosting, a managed private deployment, a provider with enterprise controls, or no AI for the sensitive part.\n\n## Short Answer\n\nStart with the data classification. If data cannot leave your environment, use local/open models or a private managed deployment. If data can be sent to a vendor under contract, compare provider controls, retention settings, audit needs, and model quality.\n\n## Decision Rules\n\n| Constraint | Recommendation |\n|------------|----------------|\n| Data cannot leave network | Local or private cloud deployment |\n| Regulated data | Vendor review plus logging/audit controls |\n| Trade secrets | Prefer private deployment or strong contractual controls |\n| Public content | Use best task model |\n| Mixed data | Split pipeline by sensitivity |\n\n## Questions an Agent Should Ask\n\n- What kind of data is involved?\n- Can data leave the organization?\n- Is retention disabled or governed?\n- Is audit logging required?\n- Are outputs used for decisions or drafts?\n- Does the user need model weights, API access, or a product UI?\n\n## Agent Workflow\n\n1. Classify sensitivity before recommending a model.\n2. If local is required, fetch `/api/v1/recommend/local.json`.\n3. If managed API is allowed, compare provider profiles.\n4. Recommend a small proof-of-control before a proof-of-quality.\n5. Include redaction, logging, and access controls in the plan.\n\n## Failure Mode\n\nThe common mistake is saying \"use an open-source model\" as if that solves privacy. Privacy depends on where the model runs, where logs go, who can access data, and how outputs are used.","content_length":2022,"generated_at":"2026-04-24"}