The right plan for every
compliance workload.

PDF, DOCX, DOC, and TXT files are supported across all tiers. Scanned PDFs and image-based documents are processed via OCR automatically. Maximum file size varies by tier — 25 MB on Ephemeral up to 250 MB on Vector and custom on Enclave.

One question submitted against a document counts as one query. If you ask three follow-up questions, that is three queries. Cache hits — identical questions on the same document — count toward your daily limit but are served instantly from the semantic cache at no additional inference cost. Unused queries do not roll over to the next day.

Yes — your document chunks are stored in a private vector index scoped to your account. PII is scrubbed before storage and never passed to any language model in raw form. You can delete any document at any time from your document history panel.

Meridian Intel routes queries through Mistral-7B local inference, then Gemini 2.5 Flash, then GPT-4o-mini as fallback. All answers are document-grounded — models receive only your document chunks, not your original files. If the document does not contain relevant information, the system refuses to answer rather than generating a response from general knowledge.

No. Your uploaded documents are used solely to answer your queries. They are not used to train or improve Meridian Intel's models, and are not shared with third parties for training purposes.

Yes. Ephemeral is a permanent free tier — 10 queries per day and 3 document uploads per week, with no time limit and no credit card required. It is not a trial that expires.

The first 10 customers who sign up on a paid tier lock in today's pricing permanently — even after prices increase. If you start on Segment at $29/mo, that rate is yours for as long as you stay on Segment, regardless of future price changes.

Yes. You can upgrade at any time. Downgrading moves you to the current rate for that tier — founding customer pricing only locks in the tier you originally signed up on. Stripe integration for self-serve billing is coming soon; contact support@meridianintel.app in the meantime.

Queries are blocked once the daily limit is reached. The counter resets at midnight UTC. On Vector and Enclave tiers, queries are unlimited. You will see your remaining count in the workspace usage panel.

Annual billing with a discount is on the roadmap. Not yet available. Contact sales@meridianintel.app if this matters for your procurement process.

Enclave is designed for regulated organizations — pharma QA teams, legal compliance functions, financial audit departments — where mandatory MFA, audit trails, and data governance are non-negotiable requirements. If you need SSO, a BAA, or an on-premise deployment path, contact sales@meridianintel.app.

On-premise means running Meridian Intel entirely within your own infrastructure — no data leaves your environment. This will be delivered as a Docker package for Enclave customers. It is currently in development for Phase 7 and not yet available. Contact sales to register interest and be notified when it launches.