What an Agent is made of
Base model+Instructions, knowledge & files+Skills, tools & integrations→Your AI AgentUsable in Chat and Flow
MTN AI · AI platform
AI for your whole company,
in one account.
Chat, images, Agents and file workspaces on models from several providers, with company-wide rules for data and spend.
Available model families
Models
Pick the model per task, not per subscription.
One account gives you models from Anthropic, OpenAI and Google plus open models hosted fully in the EU.
- Switch models partway through a chat; the conversation carries over
- Filter the model list by reasoning, vision and web search
- Open models such as DeepSeek, GLM and Kimi run on EU-hosted infrastructure
EXAMPLE MODELS
Claude
OpenAI GPT
Gemini
DeepSeek
GLM
Kimi
Logos are trademarks of their respective owners, shown for reference only.
Chat
Answers can cite the web and read your attachments.
Draft, analyze, translate, summarize and code in one conversation.
- Adjustable reasoning effort with a visible reasoning summary
- Live web search with linked sources
- Attachments: images, PDFs, Office files and audio
Files & workspaces
Give the AI a folder to work in.
A workspace is a folder from your files that you attach to a chat or a Flow.
- The AI lists, reads, searches, creates and edits the files in it, including Word and Excel
- Every edit is a new version you can compare and restore
- One workspace per chat, or one shared by every node of a Flow
- Download the whole folder as a .zip at any time
Files are opened when needed, not sent along with every message, so long chats stay fast.
AI Agents
An Agent is a model with a fixed job.
Configure it once, then pick it in Chat or Flow like any model.
- A lead Agent can hand bounded tasks to specialist Agents and combine their results
- Every handoff can be opened and checked in Chat or Flow
- Tools include web search, files, images, skills and MCP servers
A lead Agent delegates to specialists
Your requestOne clear outcome
→Lead AgentPlans, delegates, checks
→- ResearcherFind evidence
- AnalystTest the facts
- WriterShape the answer
- ReviewerCheck the work
Combined resultSynthesized by the lead
Canvas
Generate and edit images in the same account.
Images are paid from the same balance as chat, without a separate image subscription.
- Several image models for different styles
- Set quality and size per image
- Edit with reference images; results collect in one gallery
Flow
Branch one prompt and compare the results.
Flow lays a conversation out on a canvas, so one prompt can branch into several paths.
- Use a different model on each branch
- Keep every branch and continue the strongest
- Compare drafts, analyses or plans side by side
Privacy
Every model shows how its provider handles your data.
You see the tier before you send, and companies can require a minimum tier for everyone.
Privacy tiers, strongest to weakest
- European Union
Provider hosted in the EU, zero retention, no training.
- Confidential
Zero retention at the provider, no training; processing may happen outside the EU.
- Protected
Provider processes data in accordance with the GDPR; temporary retention possible, no training.
- Zero Data Retention
Zero data retention, no training.
- Basic Privacy
The provider may retain data; no training.
- Unrestricted Use
No guarantees; data may be used for training.
Tiers describe the model provider only. Your account, chats and files are hosted exclusively in the EU and processed in accordance with the GDPR.
Integrations & API
Connect the tools you already use.
Connected tools are given to Agents, tool by tool.
Included in Freelancer, Teams+ and Teams BYOK · Individual on request
Included integrations
Ready-made connectors for widely used services; more on request.
Your own MCP servers
Connect any service that speaks MCP, including your in-house servers. An admin approves every server before it is used.
Add-on · individual contract
Premium integrations
Connectors we build and maintain for systems without MCP, usually for a specific customer request.
- Credentials stored encrypted in your organization
- Optional seat licences per user or department
Add-on · individual contract
API
Use MTN AI from your own software through an OpenAI-compatible API, available for every plan on request.
- Same models, Agents and files as the web app
- Works with OpenAI SDKs and coding tools
- Requests run with the key owner’s permissions and model policies
Pricing
Monthly platform fee plus what you use.
Personal use has no platform fee; company plans add seats and controls.
Individual / Non-Commercial
For personal, non‑commercial use. Freelancer features without a platform fee, billed purely by usage; integrations and MCP servers on request.
Agent scope: Personal AI Agents
Non-profit? Contact us to put your whole organization on a Teams plan.
0€/ mo
Register nowPlans for commercial use
Pay yearly and get one month free
Freelancer
8€/ month
- Forfor 1 user
- Extra seats–
- Billing+ Usage-based billing
- AgentsPersonal AI Agents
- AI Assistant Chat
- Image Generation and Editing
- Transparent privacy
- Multi-provider model selection
- Usage and credit tracking
- Credit Balance (spend control)
- Integrations and your own MCP servers
Teams
38€/ month
- Forfor 5 users
- Extra seatsAdditional 5 users + 28€ / month
- Billing+ Usage-based billing
- AgentsAI Agents not included
- Chat, images, Canvas and Flow for 5 users
- Unified company billing
- Enforce minimum privacy tier
- Maximum spend per user
- Maximum spend per month
- Usage reporting
Teams+
48€/ month
- Forfor 5 users
- Extra seatsAdditional 5 users + 38€ / month
- Billing+ Usage-based billing
- AgentsPersonal + organization AI Agents
- Everything in Teams
- Organization AI Agents with skills and specialist Agents
- Organization MCP tool integrations
- Company-wide or department visibility
- Admin-locked Agent settings
Add-onsPremium integrations · APIIndividual contract
Commercial plans: pay yearly and get one month of the platform fee free (usage excluded). All prices exclude 19% VAT.
Usage-based billing on top
Completed responses billed through MTN credits show their billed cost by default.
For large organizations
Teams BYOK
For organizations with their own provider contracts: bring your own API keys and pay providers directly at your negotiated rates, with no token markup from us.
Agent scope: Personal + organization AI Agents
- Bring your own keys for supported AI providers
- Everything in Teams+, running on your own keys
Custom pricing
Talk to salesFAQ
Frequently asked questions
A saved prompt is text you paste again. An Agent also keeps its base model, knowledge, files, settings, skills and tools, and you pick it in Chat or Flow like a model. Built-in capabilities still depend on the selected base model.
Yes. A lead Agent can delegate bounded tasks to the specialist Agents configured for it and combine their results; billed cost is shown when applicable. Specialists do not call further specialists.
Individual and Freelancer include personal AI Agents. Teams does not include AI Agents. Teams+ and Teams BYOK include personal Agents plus organization Agents that can be shared company-wide or limited to selected departments.
An ordinary folder from your Files page that you attach to a chat or a Flow. With every message the AI gets a short list of its files and opens only what it needs. Each edit saves a new version; the last 10 older versions are kept for up to 30 days, and deleted files can be restored for 30 days. Only models with the File tools badge can work in a workspace.
An MCP server is a connector a service already offers through the Model Context Protocol: you add its address, an admin approves it, and its tools become available to your Agents. A premium integration is a connector we build for a system that has no MCP server. It is an add-on on an individual contract.
Yes, through the MTN AI API, available for every plan on request on an individual contract. It follows the OpenAI request format, so most OpenAI SDKs and tools work once you change the base URL and key. Requests run as the key owner and are billed like the web app.
Tokens are the small units AI models use to measure text: a short word is roughly one token, while longer words can be a few. Text-model charges are based on input and output tokens; images, requests and tools can use other billing units. Completed responses billed through MTN credits show their billed cost by default; BYOK usage is billed by the provider.
On every tier except Unrestricted Use, your prompts are not used for model training. How long the provider may keep them depends on the tier.
Your account, chats and files are hosted exclusively in the EU and processed in accordance with the GDPR. When you send a message, it goes to the selected model’s provider, and that model’s privacy tier applies. Contractual GDPR and deployment details are available during security review; after signing up, you can access our GDPR Data Processing Agreement (DPA) in the MTN Trust Center.
Start for free, or plan a rollout with us.
Register for personal use, or contact us about company plans, add-ons and the API.
MTN Media Group is not affiliated with, endorsed by, sponsored by, or in partnership with OpenAI, Anthropic, Google, or any other providers named on this page. All product names, logos and trademarks are the property of their respective owners and are used for identification only.