Pricing
Honest pricing for AI that ingests, retrieves, and acts on your documents.
A flat monthly subscription with three direct meters — pages, tokens, tool actions — each with a monthly allowance included. Beyond the allowance, you choose per meter whether to pause or pay the published overage rate. No credits, no surprise invoices.
Generous allowances on every meter. Overage only if you opt in past the cap. You control which meters allow overage and where the monthly ceiling sits. Pages preview cost before every import.
Starter
$99/mo
For small teams indexing their first knowledge base.
7,000 pages/month
10M tokens/month (scaled by model)
500 tool actions/month
Standard parser
Customer-facing embed included
Email support, 48h
Most popular
Pro
$199/mo
For teams running production document AI.
20,000 pages/month
20M tokens/month (scaled by model)
1,000 tool actions/month
Standard + opt-in Premium parser
Customer-facing embed included
Email + chat support, 8h
Enterprise
Custom pricing
For regulated industries and large corpora.
Custom page, token, and action allowances
Premium parser default
BAA available
Dedicated support contact
Dedicated VPC deployment available
How the meters work
Each meter, explained.
Your dashboard shows three numbers — pages used, tokens used, tool actions used — each against the monthly allowance on your tier. A page is a page, a token is a token, an action is one tool call. Below: what each meter measures and what happens when you reach the cap.
/01
Default: pause
Pages
What it measures
Document and web ingestion into your knowledge base. Counted per consumed page at the time of import.
At the cap
Ingestion pauses past the allowance — admin-driven work can wait. Switch to "allow overage" at $0.03/page and every import shows the page count and cost on a Confirm/Cancel preview before any billing.
/02
Default: allow overage
Tokens
What it measures
Input + output tokens consumed by chat queries and agent reasoning. Scaled by the model you choose — see the per-model multiplier table below.
At the cap
Chat and agent reasoning continue past the cap, billed at $15 per 1M tokens (baseline multiplier). End users keep getting answers. The monthly overage ceiling caps total spend. Switch to "pause at cap" if you prefer a hard wall.
/03
Default: allow overage
Tool actions
What it measures
Agent tool invocations through Composio, MCP, or OpenAPI integrations. One action = one tool call.
At the cap
Agent tool calls continue past the cap, billed at $0.30 per action. Time-sensitive workflows don't stall. The monthly overage ceiling caps total spend. Switch to "pause at cap" if you prefer a hard wall.
Skills and Agents — which the tokens and tool-actions meters power — are in early access. Production-usable today, actively iterating. The metering and overage controls described above apply now; surface details may shift while the product matures.
Monthly overage ceiling
Set an optional monthly overage ceiling in admin and every overage-enabled meter auto-pauses once it's reached. That's your seatbelt: a runaway agent loop that would have generated $4,000 in token overage stops at your ceiling instead, with an email notification. No surprise invoices — by design, not by marketing.
Per-model multiplier
Tokens scale with the model you choose.
Every supported AI model has a published multiplier relative to the Haiku 4.5 baseline. When you use a more expensive model, your token budget depletes faster — because that's the genuine cost shape. Use a cheaper model and you get more queries from the same budget. You can switch models per agent at any time.
Model
Multiplier
Nova Micro / Mistral Small
0.1×
Haiku 4.5 / Nova Lite
1× (baseline)
GPT-5 nano / mini
1.5×
Sonnet 4 / GPT-5
6×
Opus / GPT-5 Pro
15×
Worked example
A 1,000-input + 200-output query on Sonnet 4 deducts (1,000 + 200) × 6 = 7,200 tokens from your monthly budget. The same query on Haiku 4.5 deducts 1,200.
The multiplier is published; you can predict exactly how your budget will deplete for any model choice.
How import pricing works
Preview every import. Pay only what you confirm.
When you add documents to your knowledge base — or point Dokko at a website — we process them once. Reading, understanding structure and content, indexing for retrieval. You're charged per page at that point, not on every query. Your plan includes a monthly page allowance (7,000 on Starter, 20,000 on Pro, custom on Enterprise); beyond it, the published $0.03/page overage rate applies.
Before every import, Dokko shows you the page count, the parser choice, and the cost. You confirm or cancel. A stated cost you approve is a cost you will never dispute.
Process Knowledge Base: Compliance Archive
Every import. Every time.
Parser choice
Standard or Premium — you choose per knowledge base.
Vision-capable
Standard parser
Handles most PDFs well at low cost
Routine contracts, modern briefs, clean text-based documents
Counts 1× from your page allowance
Available on every tier
Higher-fidelity vision
Premium parser
Best-in-class on dense, complex documents
Borderless tables, multi-column layouts, regulatory archives, scanned contracts, financial filings
Counts 5× from your page allowance (reflects the higher processing cost)
Not available on Starter; opt-in per knowledge base on Pro; default-on for Enterprise
The 5× weight is the only place quality choice has a cost implication, and it's shown to you before every import. Non-PDF files (Word, spreadsheets, presentations, plain text, web pages) always count 1× — parser choice only affects PDFs.
Pricing FAQ
For PDFs, one page = one page of the document. For everything else (DOCX, spreadsheets, presentations, Markdown, plain text, CSV, web pages), one page = roughly 2,000 characters of extracted text. Web pages are measured after navigation, headers, and ads are stripped. Premium PDFs use 5× from your allowance to reflect their higher processing cost.
Standard handles most PDFs at low cost — routine contracts, modern briefs, clean text-based documents. Premium is significantly better at dense financial documents, regulatory archives, and borderless tables — at significantly higher processing cost. Starter uses Standard only. Pro lets you opt into Premium per knowledge base. Enterprise uses Premium by default.
Tokens are something you can verify — every model SDK reports them. We publish the per-model multiplier so you can pick the right model for each job and see exactly how your budget depletes. Credits hide the underlying cost shape; tokens with a multiplier transmit it honestly.
That's your choice per meter. By default, tokens and actions allow overage at the published rate (so end users aren't blocked mid-month) and pages pause past the allowance (ingestion is admin-driven; waiting is fine). In admin settings you can switch each meter independently. Token overage is $15 per 1M tokens (baseline multiplier); action overage is $0.30 per action; pages overage is $0.03 per consumed page with a Confirm/Cancel preview before every import. Set a monthly overage ceiling and all overage-enabled meters auto-pause once it's reached — your seatbelt against runaway agent loops.
No. Your subscription covers all chat and agent queries up to the monthly token allowance. Different models deplete the token budget at different rates (see the multiplier table). Beyond the allowance, behaviour depends on your token cap policy.
All plans include access to every supported model across major providers (Anthropic, Amazon, Meta, Mistral, OpenAI, Google, Groq). You assign specific models per agent from the dashboard. Use cheaper models for routine queries and premium models only where they earn their cost.
Document ingestion costs vary by orders of magnitude — a 10-page contract and a 5,000-page regulatory archive cost dramatically different amounts to process. Flat-rate vendors handle this two ways: hide the cost inside per-seat prices that only work at $1,000+/seat, or absorb thin allowances and surprise you with mid-cycle upsells. Our plans include a generous monthly allowance on every meter (pages, tokens, actions). If you genuinely need more, overage at the published rate is opt-in and previewed before any charge.
Yes, on every tier. Chat tokens routed through your own provider key are billed by your provider — they don't count against your Dokko token allowance and Dokko doesn't mark them up. Ingestion, reranking, and tool actions still run on Dokko's infrastructure.
Yes. 14 days at full Starter capability (7,000 pages, 10M tokens, 500 actions, customer-facing embed included, BYOK supported). Credit card required at registration. No charge during the trial; cancel any time.
Because SSO isn't shipped yet. When it ships, it will be listed where it belongs (Enterprise) and not before.
Enterprise includes custom limits, dedicated infrastructure options, BAA, dedicated support contact, white-glove onboarding, and multi-year price lock. Talk to sales.