Dashboard pricing
Bring your own model and enjoy one flat price for observability and hosting.
Starter
For building and trying the Dashboard in development.
- 1 seat
- No live projects, development only
- 1 local project
- SOC 2 Type II compliant
- Coparlant, 100 messages / month
- Support on Discord
What you can view and analyze
- Monitor local development sessions
- Session history: 100 sessions or 7 days
Basic
RecommendedFor your first agent in production.
- 1 seat
- 1 live project, managed or self-hosted
- 3 local projects
- SOC 2 Type II compliant
- Coparlant, 250 messages / month
- Other AI agents, such as ChatGPT, Instinct, or Muse, talking to yours on someone's behalf.OpenAI dotsMeta MuseGrok BotInstinctClaude Duet: 5 messages / session, 200 messages / day
- Support on Discord
Meta Muse
Grok Bot
Instinct
Claude DuetWhat you can view and analyze
- Monitor up to 1,000 sessions / month
- Session history: 1,000 sessions or 30 days
Plus
For teams running several agents.
- 5 seats
- 3 live projects, managed or self-hosted
- 10 local projects
- SOC 2 Type II compliant
- Coparlant, 1,000 messages / month
- Other AI agents, such as ChatGPT, Instinct, or Muse, talking to yours on someone's behalf.OpenAI dotsMeta MuseGrok BotInstinctClaude Duet: 10 messages / session, 1,000 messages / day
- Support on Discord
What you can view and analyze
- Monitor up to 5,000 sessions / month
- Session history: 15,000 sessions or 90 days
Prices in USD. Plans never limit how many conversations your agents have: session limits are what the Dashboard ingests for you to view and analyze. Each plan applies to one workspace, and you can run several workspaces on different plans. You bring your own LLM; inference is never marked up.
Questions.
A plan applies to a workspace. If you need different plans for different teams or clients, you can manage several workspaces, each on its own plan.
A live project is an agent in production, either hosted by us or self-hosted on your own servers and connected to the Dashboard, which ingests its sessions up to your plan's monthly limit. A local project connects your local development environment, so you can use the Dashboard while you build.
Not with your customers. Your agents talk to as many customers as they need to, on every plan. The session limits only set how many conversations the Dashboard ingests each month for you to view, trace, and analyze, and the retention limits set how long it keeps them. The one exception is other AI agents talking to yours over A2A: there, your plan sets how many replies each conversation and each day can have.
Starter is for development. Connect your local environment to try out the Dashboard with one local project. When you're ready to put an agent in front of real customers, move to Basic or above.
Every plan's yearly price is ten times its monthly price, so paying for a year gives you two months free.
No. You bring your own inference: OpenRouter, OpenAI, Gemini, Anthropic, whichever provider you already use, with your own key. Your model bill stays between you and your provider; we never sit in the middle and never mark it up.
Yes. Parlant is Apache-2.0 licensed and always will be. The Dashboard is an optional layer on top: you can build, run, and ship agents on the open-source framework without ever creating a Dashboard account.