An AI platform comparison built around your first need: a website chatbot, connected agents or a shared knowledge base.
Values separated by “/” follow the order of the plans listed. Dollars are US dollars; no conversion to euros is applied.
On a small screen, scroll the table sideways. The criteria stay visible on the left.
| Your criteria | Narratheque | Chatbase | Dify | Dust | Mistral |
|---|---|---|---|---|---|
| Plans compared | Simple / Pro / Max · proposed | Hobby / Standard | Professional / Team | Business · Pro seat | Team |
| Monthly price | €29 / €129 / €249 excl. VAT | US$40 / US$150 | US$59 / US$159 excl. tax | €30 excl. VAT / user | US$24.99 excl. tax / user |
| Billing unit | Organisation + extra bases and seats | Plan + extra agents | Workspace | User seat | User |
| Users included | 2 / 10 / 30 | 2 / 3 | 3 / 50 | By seats subscribed | By seats subscribed |
| AI budget included per month | 50,000 / 250,000 / 750,000 NrTok (€5 / €25 / €75) | 700 / 4,000 credits | 5,000 / 10,000 credits | 8,000 credits per seat | Fair-use limits |
| Bases or applications included | 1 / 5 / 15 bases | 1 agent | 50 / 200 applications | Custom agents | Collaborative space |
| Choice of several LLM providers | From Simple, within the models available | Several providers announced | Several providers and API keys | Several providers announced | Mistral ecosystem |
| Putting your content to work | Vector RAG, wiki and protection from Simple | Content supplied to the agents | RAG bases and workflows | Knowledge and connected tools | Documents and libraries |
| Publishing chatbots | 1 / 5 / 15 chatbots | Yes, customer relations agent | Applications and chatbots | Public scope to be confirmed | Public scope to be confirmed |
| Synchronised sources | Within the catalogue and volumes confirmed; processing in NrTok | Automatic refresh announced on Standard | Knowledge pipelines | Connectors to tools | Connectors announced, scope to be checked |
| Identifier protection before processing | No extra charge per base; processing in NrTok | Comparable mechanism not verified | Comparable mechanism not verified | Comparable mechanism not verified | Comparable mechanism not verified |
| Oversight and governance | Shared access and budget; governance to be framed | SSO and audit on Enterprise | Role management | SSO; advanced audit on Enterprise | Audit and SSO on Enterprise |
| What comes on top of the price | NrTok, capacity and specific integrations | Usage and agents | API usage and resource limits | Seats and overages | Usage and specific deployments |
| Source and scope | Proposed pricing ↗ | Source: Chatbase ↗ | Source: Dify ↗ | Source: Dust ↗ | Source: Mistral ↗ |
A credit is not worth the same from one platform to another. The amounts compared are subscriptions, not total costs at equivalent usage. Features, limits and services differ. “Not verified” means information this comparison has not established, not the absence of a feature.
Narratheque presents its newly proposed pricing here, with ceilings and terms still to be confirmed. Competitors are described from their public offers. Any Chatbase taxes are to be checked at payment.
Our reading of the offers: these pointers help you draw up a shortlist; they do not replace a trial on your own documents.
Chatbase is worth looking at for customer relations and its dedicated features. Narratheque Simple offers a way in at €29 with a base you can reuse for future internal needs. Compare the volume of answers and the support features too.
Dust is worth looking at for its agents and its connections to tools. Narratheque Pro offers 5 bases and 10 users for €129. The choice depends on the integrations available and on your usage.
Dify is worth looking at for its workflows and its building environment. Narratheque favours a path around bases, assistants and chatbots; specific integrations are to be scoped.
5 bases · 10 users
250,000 NrTok shared (€25)
10 × 30 €
8,000 credits per seat
Up to 50 members
10,000 credits monthly
An arithmetic example, with no negotiated discount or annual billing. These plans do not deliver an equivalent volume of work. Dust also allows different seat types to be combined. To compare the real cost, test the same documents and tasks, then add usage and services.
Ask where storage and processing happen: documents, index, requests, extracts and logs. Examine the path through to the model selected.
Check access by team, the separation of bases and what a chatbot publishes. Have those rights tested on your real case.
Ask about the training, retention and deletion terms that apply to each model provider and to your configuration.
Check the exports, changing model, the connections available and the limits on circulating answers drawn from your knowledge.
Dust announces data residency options in the United States and the European Union. Chatbase reserves SSO, audit logs and zero data retention, among other things, for its Enterprise plan. These statements have to be pinned down contractually for your configuration. Source: Dust ↗ Source: Chatbase ↗
For Narratheque, the processing path, the models available and the scope of protection are to be confirmed before handling sensitive data. This comparison establishes no ranking on compliance or security.
One base can feed an internal assistant and the applications you authorise. You add new scopes when your teams or your audiences call for them.
RAG document search, wiki organisation and protection of identifying information answer different needs. The wiki comes from Pro onwards. Advanced protection costs €20 more per base; Max includes three. Processing consumes credits.
An agency or an AI lead can help you prepare the sources and roll out your first use case. That support is quoted separately from the subscription.
Quality depends on your documents, the models, the configuration and the tasks. We have run no common performance test; the bearings given here rest on the official pages consulted.
Not necessarily. Compare the subscription, the AI usage, preparing the sources, the connectors and the support on one and the same scenario.
No. This selection covers chatbot, application-building and team AI offers. Free editions, annual discounts and negotiated contracts are not compared in detail.
Yes. Prepare a small authorised corpus and representative questions. Judge the quality and the sources of the answers, the access rights, the preparation time and the real cost. Avoid introducing sensitive data before the processing path has been approved.
An editorial comparison published by Narratheque, itself a player in this market. Competitors’ prices checked on . Narratheque pricing proposed on 14 September 2026. Competitor sources: the official pricing pages below. Prices may change and must be confirmed with each publisher.
The descriptions of capabilities repeat what is announced; the usage recommendations are our own analysis. No comparative trial of speed, accuracy or security is claimed.
One base, one audience and a few tasks: the right bearings for choosing.
Obtenez des réponses instantanées à vos questions sur la base des contenus du site internet
