Pricing

Pay per agent, not per person

Start free for 14 days. Then pay for the agents you deploy, with runs included and model usage at cost or on your own key. Hosted by us, or self-hosted under licence.

Plans

Choose a plan

List prices, effective 6 November 2026. US dollars, or Indian rupees for companies in India, before taxes. A quote or order form may differ, and the order form governs.

Free trial

Try the platform on your own work before you pay anything.

Freefor 14 days

No card needed. Nothing is charged.

Start free

Includes

  • A hosted workspace of your own
  • Up to 3 agents and 1,000 runs
  • The five pre-built solutions on sample systems
  • US$10 (₹800) of model usage, or your own key

Team

One team putting its first agents to work.

US$99per deployed agent per month

₹7,999 in India. Or US$83 (₹6,666) a month billed yearly: two months free.

Start free

Everything to start

  • Up to 10 deployed agents; drafts are free
  • 2,000 runs per agent a month, pooled
  • All 50 connectors, plus MCP, OpenAPI and OData
  • Approvals, budgets, guardrails and kill switches
  • Single sign-on with Google, Microsoft or OIDC

Business

Agents across several teams, under your company's identity and policies.

Custombilled monthly or annually

Volume pricing for agents and runs.

Contact sales

Everything in Team, plus

  • No limit on agents
  • SCIM 2.0 provisioning from your directory
  • Cedar policies for every person and agent
  • A one-year, verified audit log
  • Model calls held to the region you name

Enterprise self-hosted

The whole platform in your own environment, under licence.

Customannual licence

Runs in your own cloud or data centre.

Contact sales

Everything in Business, plus

  • No run limits on your own infrastructure
  • Docker Compose, or Kubernetes with the Helm chart
  • Your own models, including Gemini on Vertex AI
  • Updates for the licence term
  • A named support contact

Compare

Compare plans

Free trialTeamBusinessEnterprise self-hosted
Where it runsHosted by LiyagentHosted by LiyagentHosted by LiyagentYour cloud or data centre
Deployed agents3Up to 10No limitNo limit
Runs included1,0002,000 per agent a month, pooledVolume pricingNo limit
Extra runs-US$0.02 (₹1.50) eachVolume pricing-
Connectors, MCP, OpenAPI, ODataYesYesYesYes
Approvals, budgets, guardrailsYesYesYesYes
Single sign-on (Google, Microsoft, OIDC)-YesYesYes
SCIM 2.0 provisioning--YesYes
Cedar policies--YesYes
Audit log keptFor the trial90 daysOne year, verifiedYou decide
Model calls held to a region--YesYes
Model usageUS$10 included, or your keyYour key, or at costYour key, or at costYour own models
SupportEmailEmail, within one working dayPriority, within four working hoursNamed contact

Model usage

The model is yours to choose

Bring your own key

Gemini, Vertex AI, Anthropic, OpenAI, Amazon Bedrock or any OpenAI-compatible endpoint. You pay the provider directly.

Or use ours, at cost

Model calls through Liyagent's own accounts are billed at the provider's list price, with no markup.

Capped before every call

Each agent has a budget, checked before each model call, so a runaway loop stops at the limit you set.

Get started

Three ways to start

  1. 01

    Start free

    A 14-day trial on a hosted workspace of your own. No card needed.

    Start free
  2. 02

    Try it live

    Ask Liya to work in five sample systems right now, no sign-up.

    Try it live
  3. 03

    Book a demo

    Thirty minutes on your use case, with the systems you run.

    Book a demo

Questions

Pricing questions

What counts as a deployed agent?

An AI agent switched on for people or systems to use: a pre-built one such as Liya Triage, or one you built. Agents in draft or switched off are free, and sub-agents inside an agent are not counted. It is an AI agent, not a person: people who talk to your agents are never charged.

What is a run?

One task an agent carries out from start to finish, however many steps it takes: one request in chat, one ticket triaged or one scheduled job. Included runs are pooled across your workspace, and extra runs are billed monthly in arrears at the list rate.

Who pays for the AI model?

You choose. Bring your own key for Gemini, Vertex AI, Anthropic, OpenAI, Bedrock or any OpenAI-compatible endpoint and pay that provider directly, or use Liyagent's model accounts, billed at the provider's list price with no markup, separately from runs. An agent that reaches its budget pauses, so there is no surprise bill.

Is there a charge per user?

No. Everyone in your company can talk to your agents in Teams, Slack, Google Chat, email or the web app. You pay for agents and runs.

What happens when the trial ends?

Nothing is charged. Choose a plan to keep your workspace and agents, or it is deleted 30 days after the trial ends.

How do we pay?

Monthly, or yearly with two months free, by invoice in US dollars, or in Indian rupees for companies in India. Prices exclude taxes, and Indian prices exclude GST. Business and Enterprise are quoted by our sales team.

Can we self-host?

Yes, under the Enterprise licence: Docker Compose on one VM or Kubernetes with the Helm chart, in your own cloud account (Google Cloud, AWS or Azure) or your data centre. The source code is not part of the licence.

Liyagent is licensed software, sold as a subscription or an annual licence. Its source code is not part of any plan, and the liya CLI you can download carries none of it.

Put agents to work where your business runs