Plans & pricing
We don't count messages. Your plan sets the weekly limit for model work and what the agent does without you.
Pro
PopularFor everyday work: reports, orders, documents
- The agent’s own cloud computer
- Top models
- A limit for everyday work
- Scheduled tasks
- 110+ services: 1C, CRM, marketplaces
Max
For heavy workloads, when the agent works all day
- 5× the Pro limit
- All top models
- No queue at peak hours
- Video from a description
- Telegram support replies within an hour
Team
For organizations: Pro or Max for every member
- Every member gets their own weekly limit
- One invoice and contract for the organization
- Shared drive and knowledge base
- Member roles and permissions
- Log of every action in the organization
Compare features across plans
Integrations, search and memory are identical on every plan. The weekly model limit, the work done without you and the organization features are what differ.
| Features | Pro | Max | TeamPro or Max per person |
|---|---|---|---|
| Usage | |||
| Weekly limit | A limit for everyday work | 5× the Pro limit | Its own weekly limit on every seat |
| Requests on GPT-6 Luna | ≈ 104,456 | ≈ 522,284 | Its own weekly limit on every seat |
| Requests on GPT-6.1 Sol | ≈ 5,292 | ≈ 26,464 | Its own weekly limit on every seat |
| Requests on Claude Opus 5.5 | ≈ 13,232 | Its own weekly limit on every seat | |
| Content generation | |||
| Document generation | |||
| Images per day | Up to 45 | Up to 110 | Its own weekly limit on every seat |
| Videos per month | Up to 30 | Same as the seat's plan | |
| Integrations and tools | |||
| CRM and marketplaces | |||
| Web search | |||
| Legal database | |||
| Calculations and table analysis | |||
| Scheduled tasks and triggers | |||
| Memory of projects and clients | |||
| Organization features | |||
| Unified billing | |||
| Shared organization drive | |||
| Roles and access | |||
| Audit log | |||
| Up to 150 paid seats | |||
| Support | |||
| Email support | |||
| Refund for unused time | |||
Estimated for a request with 15k input tokens (4k from cache) and 1k output tokens. Cache costs about a tenth of fresh input, which is why the counts run this high. Images and video draw on the same weekly limit.
A request or task run already under way finishes. Once the weekly limit is reached, the app stops sending new messages, in a messenger the agent keeps answering on GPT-6 Luna, and scheduled task runs wait for the limit to reset. The limit can be reset sooner, once a month, for half the plan price. Task runs draw from the same limit and are also capped per day and per month by the plan.
Pricing FAQ
Working every day, need GPT-6.1 Sol, scheduled tasks and images: Pro. Hitting the weekly limit, need Claude Opus 5.5 and video: Max. Working with your staff: the Team plan, the same Pro or Max across as many paid seats as you need. You can start with a week of Pro, no card needed.
Every plan carries a weekly limit for model work, with no monthly cap. Chat, scheduled tasks and generation all draw from the same limit. How far it goes depends on the length of the chat, the complexity of the tasks and the tools you connect, which is why there is no fixed message count. Once it is reached you can wait for it to reset, move up a plan or, on Pro and Max, buy a limit reset. The limit is checked when a request starts: a request or task run that began before it ran out finishes. Once it is reached, the app stops sending new messages, in a messenger the agent keeps answering on GPT-6 Luna, and scheduled task runs wait for the limit to reset. Scheduled task runs draw from the same weekly limit and are also capped per day and per month by the plan. Some connections (currently Wildberries) also draw from the limit for their API calls; those calls show as their own line under Usage. Your own usage is in settings, under Usage. Some models are available only on higher plans; the table below shows which.
Pro and Max start at 20 $/mo billed monthly. Paying yearly takes 17% off. For an organization the same plan is multiplied by the number of paid seats.
Pro is for individual use: it includes all core features and the main GPT-6.1 Sol model, with document analysis, CRM and marketplace integrations, AI analytics. Max is for intensive daily work: much higher request limits, the strongest model, Claude Opus 5.5, for long documents and complex analytics, and video from a description.
Yes. The Team plan is Pro or Max across the number of paid seats you set, up to 150. A paid seat gives that member the full plan. Seats can be added any time, charged by the day for the rest of the period, and dropped seats go at renewal.
Yes. Within 14 days of payment, a full refund if no more than 10% of your plan volume was used; after that, proportional to the unused remainder. Contact support and we'll refund within 10 business days to the original payment method.
Yes. Connect your OpenAI, Anthropic, or YandexGPT key in settings, and model requests go through it directly. Integrations, skills, tasks, and memory continue to run on your plan.
SSO, SLA, GDPR-compliant servers in the EU/EEA and on-premise for companies with special requirements