Chatwoot Review 2026: Pricing, Cloud vs Self-Host

Cloud lists Hacker at $0 for two agents and 500 conversations/month, with live chat and thirty-day retention. Startups is $19 per agent/month with a help center, a 300-credit shared AI allowance, and one-year listed retention. Business is $39 with voice, Teams, automation rules, 500 credits, and two years; Enterprise is $99 with SSO/SAML, audit logs, 800 credits, and three years as listed, subject to contract. Paid conversation volume remains subject to the fair-use policy, so 'unlimited' is not a promise of unlimited infrastructure or AI processing.

Captain allowances belong to the account, not each paid agent. The credit guide charges by AI actions or messages; several replies and a summary can consume more than one credit in a single customer conversation. Extra packs are listed at $20 per 1,000 credits, and the billing explanation describes a six-month life for purchased credits. Check these rules before treating a credit as a resolution or multiplying an allowance by seat count.

Self-hosting gives more control over infrastructure, but the edition matrix must be evaluated separately from Cloud. Community excludes Captain, voice, custom branding, granular roles, SSO, and SLA policies. Paid self-hosted editions unlock different capabilities: Premium Support includes Captain, while Enterprise adds controls such as SSO and SLAs. The similarly named software edition and commercial subscription tiers should not be conflated. A downloadable codebase is not proof that every Cloud feature is free to run.

Captain's self-hosted setup guide requires a paid Enterprise-edition installation and model credentials. If customer content is sent to an external AI provider, hosting the inbox yourself does not keep all processing inside your infrastructure. Add license costs where applicable, model usage, backups, patching, monitoring, and restore ownership to the comparison. Cloud may reduce operational work; self-hosting may satisfy a specific infrastructure requirement. Compare both with Crisp or Tidio against a real set of needs rather than making a universal cheapest-platform claim.

Feature-wise, the documented set covers what a startup support operation needs: omnichannel inbox, help center from the first paid tier, automation rules and teams at Business, and APIs for export and integration. The gaps are ecosystem-shaped rather than feature-shaped: Zendesk's marketplace depth, Intercom's Fin AI maturity, and the admin tooling of incumbent suites remain ahead, and retention caps (1-3 years by tier) may matter to compliance teams.

The AI comparison needs a common definition of useful work: a chargeable action is not equivalent to an accurately resolved customer issue. Measure escalation, repeat contacts, incorrect answers, and the agent time needed to finish a conversation in a controlled pilot. Our AI chatbots for customer service guide compares different charging models. For email-first support, also evaluate Freshdesk and Help Scout without assuming AI replaces routing, knowledge maintenance, or human judgment.

Against Intercom: Intercom is the category's AI flagship with per-resolution Fin pricing and a deep platform; Chatwoot counters with open code, flat rates, and self-hosting. Against Zendesk: Zendesk wins enterprise workflow depth, marketplace, and governance maturity; Chatwoot wins cost ceiling and control - our Zendesk alternatives guide frames when that trade favors challengers.

Against Crisp: Crisp's flat workspace pricing suits small teams that want messaging simplicity without open-source operations; Chatwoot's per-agent model scales differently and adds the self-host option. Against Tidio: Tidio leads on no-code chatbot building for small ecommerce; Chatwoot leads on platform ownership. And for support connected to a CRM record, HubSpot Service Hub remains the suite play in this category.

Consider a five-agent Startups account as an illustrative budget, not a quote. At the displayed rate, seats cost $95 per month. The included Captain allowance remains 300 credits, not 1,500. If five hundred conversations each trigger four chargeable AI actions, the example requires 2,000 credits. With no existing purchased balance, the 1,700-credit shortfall would require two 1,000-credit packs at the listed $20 pack price. The result is $95 plus $40 for those packs, before other applicable charges. Unused purchased credits and their expiry must be tracked separately.

Now check whether Startups actually meets the workflow. A need for Teams, automation rules, or voice points to Business irrespective of whether AI top-ups look inexpensive. Conversely, do not upgrade solely to obtain a larger allowance without pricing the alternative. Pilot representative questions, multi-message conversations, summaries, and human handoffs. Record the enabled actions behind actual consumption so the forecast can be improved. The objective is not the lowest theoretical cost per ticket; it is a budget that covers useful, safe support during a busy week.

Before routing live traffic to an assistant, test a known answer, a missing answer, a misleading customer request, and a case that must reach a person. Check what happens if the credit balance is insufficient or an external service fails. Agents should receive enough conversation context to continue rather than asking the customer to repeat everything. These are proposed acceptance checks; the review's documentary score does not measure response accuracy or uptime.

Retention also needs an operational plan. Export representative conversations, contacts, and the attachments or links the business must keep, then confirm they remain understandable outside the application. An API-export promise is not by itself a complete archive or a restorable backup. A self-hosted deployment additionally needs database and file recovery procedures with an assigned owner. Review contractual retention and regional processing requirements rather than assuming either Cloud's default settings or a chosen server location settles them. The same questions should be asked of Zendesk and Intercom to make the comparison fair.

Chatwoot earns an 8.6/10 editorial fit score as the credible open-source pillar of this category: a usable free cloud tier, flat $19-$99 per-agent pricing with published AI metering, documented export paths, and a self-hosting escape hatch no closed rival offers. The counterweights: US-hosted cloud data, retention caps by tier, governance features held to the $99 tier, a younger ecosystem, and self-hosting's real operational bill. This is a documentation-based assessment - we have not deployed it or load-tested support volumes.

Choose Chatwoot Cloud when flat pricing and openness matter and 1-3 year retention suffices; self-host when data control is a mandate and ops capacity exists. Choose Intercom for the most advanced AI resolution economics, Zendesk for enterprise process depth, and Crisp or Tidio for the simplest small-team chat. Pilot it with the free tier and one real support channel - the 15-day trial of paid features will tell you whether the credit allowance and workflows fit before any commitment.