Pricing

Pricing that scales with your corpus, not your anxiety.

Start free. Upgrade when retrieval quality becomes a team discipline. Every plan includes the full chunking pipeline.

Starter

Plan

For solo builders shipping their first RAG app.

$0/forever
Start free
  • Up to 1,000 pages / month
  • Layout-aware + semantic splitting
  • Parent-child chunking
  • Diagnostic panel (read-only)
  • 1 vector store export target
  • Community support
Most popular

Team

Plan

For teams iterating on production retrieval.

$49/seat / month
Start 14-day trial
  • Up to 250,000 pages / month
  • Everything in Starter, plus:
  • Strategy diff & embedding map
  • Retrieval replay with score breakdowns
  • Unlimited export targets
  • Versioned chunking policies
  • Priority email support

Enterprise

Plan

For organizations with scale, security, and SLA needs.

Custom
Talk to sales
  • Unlimited pages
  • Everything in Team, plus:
  • SSO / SAML & SCIM
  • Role-aware metadata & ABAC
  • Dedicated deployment & VPC peering
  • Audit logs & DPA
  • 99.9% uptime SLA
  • Dedicated solutions engineer

All prices in USD. Volume discounts available on Team and Enterprise.

What every plan includes

Layout-aware segmentation
Semantic boundary detection
Parent-child chunking
Metadata injection
Strategy diff
Embedding distribution map
Retrieval replay
Versioned policies

Pricing, plainly answered.

Still unsure? Reach out — we'll figure out the right fit together.

Talk to us
How is a 'page' counted?

A page is roughly 500 tokens of source content after OCR and normalization. Tables, figures, and code blocks count proportionally to their token footprint.

Can I switch plans later?

Yes. Upgrades take effect immediately; downgrades take effect at the next billing cycle. No lock-in, no penalties.

Do you support on-premise or VPC deployment?

The Enterprise plan supports dedicated cloud deployments and VPC peering. Air-gapped on-premise is available for select customers — talk to sales.

Which vector stores can I export to?

Pinecone, Weaviate, Qdrant, pgvector, Chroma, Milvus, and any OpenAI-compatible embeddings endpoint. Custom targets available on Enterprise.

Is my data used to train models?

Never. Your documents and embeddings are yours. We do not train on customer data under any plan.

Try CHUNKZA on your corpus

Start free. Run the diagnostic panel on your own documents. Decide when you're ready.

Start freeTalk to sales