Priced on what you actually use

A monthly platform fee with a generous allotment of processing, search, storage and seats. Go past an allotment and it bills at a published rate — it never blocks you. Every plan includes the full API.

Every plan starts with a 14-day trial · billed monthly

Standard

$374/month

For a team putting its first agentic workflows into production.

3 workspaces
10 workflows per workspace
Visual workflow builder
Semantic search — standard tier
Base generative model
Full REST API and OpenAPI spec
Standard support

Included each month

Processing units: 200
Search queries: 1,000
Storage: 10 GiB
User seats: 5

Request access

Professional

Enterprise

$9,500/month

For organisations standardising on dotDash across many teams.

Unlimited workspaces and workflows
Everything in Professional, plus
Semantic search — advanced tier with reranking
Agentic dashboards on a hosted warehouse
Data ingestion pipelines (up to 10)
Unlimited user seats
24/7 support with an SLA and a dedicated CSM

Included each month

Processing units: 5,000
Search queries: 50,000
Storage: 250 GiB
User seats: Unlimited

Talk to sales

Billed monthly on a one-year commitment.

Compare the plans

Feature	Standard	Professional	Enterprise
Workspaces	3	10	Unlimited
Workflows per workspace	10	50	Unlimited
Visual workflow builder
Semantic search	Standard	Standard	Advanced
Generative answers
Generative models	Base	Multiple	Multiple + advanced
External data integrations
Agentic dashboards
Data ingestion pipelines			10
REST API and OpenAPI spec
Standard support
Priority support
24/7 support with SLA
Dedicated CSM

Usage beyond what is included

Metered usage never blocks your work. Going over an allotment adds to that month's invoice at the rate below — it does not stop a workflow mid-run.

Dimension	Standard	Professional	Enterprise
Processing unit (per unit) Generative work: agent steps, model calls, document processing.	$1.00	$0.88	$0.63
Search query (per 1,000) Semantic, keyword and hybrid searches against your indexes.	$2.50	$1.88	$1.25
Storage (per GiB-month) Documents, embeddings and conversation history at rest.	$5.00	$4.38	$3.75
User seat (per seat) Active users beyond the included allotment.	$38.00	$31.00	Unlimited

Enterprise plans that use agentic dashboards also meter the hosted warehouse — queries scanned and warehouse storage — which is quoted as part of the contract.

What is a processing unit?

A processing unit is the platform's measure of generative work. It is derived from the tokens a model reads and writes, weighted by how expensive that model's tier is — so a small, fast model stretches your included units much further than a frontier one.

Model tiers

Every model the platform calls belongs to a tier, and the tier sets the weight. Output tokens cost more than input tokens because they cost more to produce.

Tier	Example models	InputPU per 1,000 tokens	OutputPU per 1,000 tokens
Small / fast Classification, extraction and routing — the bulk of most workflows.	Gemini 2.5 Flash, Claude Haiku	0.5	1
Standard The default for agent reasoning and grounded answers.	Gemini 3.1 Pro, Claude Sonnet	1	2
Frontier The heaviest reasoning models, for work the standard tier cannot carry.	Claude Opus	4	8
Embedding Indexing and search embeddings. These read tokens and write none, so only input is weighted.	Gemini Embedding, Text Embedding	0.1	—

How it is calculated

PU = (input tokens x input weight + output tokens x output weight) / 1,000

Every model call is measured this way, and the results are summed over your billing period.

Worked example

A grounded answer over a few pages of retrieved context on the Standard tier reads 4,000 input tokens and writes 800:

(4,000 × 1 + 800 × 2) ÷ 1,000 = 5.6 PU

Embeddings consume input tokens only, and their token counts are estimated from the length of the input — embedding models report no usage of their own. Estimated usage is recorded as such.
Indexing a document charges once. Re-indexing material that has not changed is recognised by its content and not charged again.
A model with no tier configured is metered at the standard tier, so nothing runs unmetered.
Processing units never block a run. Going past your included units adds to that month's invoice; it does not stop a workflow mid-execution.
Usage is visible in-product as it accrues, broken down by dimension, with the runs that produced it.

Models are named as examples of a tier, not as a commitment. Which tier a model sits in is configuration rather than code, and we adjust it as providers change their prices and retire or release models — the tier weights above can stay the same while a model moves between them. The weights in force are always the ones on this page.