Officework
For writing, email, research and summaries
€29.00 net · per seat/month
€34.51 incl. VAT · final price for customers in Germany
- Unlimited tokens
- Context window up to 500,000 tokens
- 1 parallel request
- Weldall Desktop Client
SeibertGPT for teams
For teams of 3 or more. More models, mix plans freely, one invoice for everyone.
For writing, email, research and summaries
€29.00 net · per seat/month
€34.51 incl. VAT · final price for customers in Germany
For agents, parallel tasks and occasional coding
€139.00 net · per seat/month
€165.41 incl. VAT · final price for customers in Germany
For coding agents, pipelines and always-on automation
€299.00 net · per seat/month
€355.81 incl. VAT · final price for customers in Germany
From 3 seats in total, whichever plan. All prices net, monthly or yearly by quote.
| Feature | Officework | Advanced | Coding & Automation |
|---|---|---|---|
| Price per seat/month, net | €29.00 | €139.00 | €299.00 |
| Tokens | unlimited | unlimited | unlimited |
| Context window | up to 500,000 tokens | up to 1,000,000 tokens | up to 1,000,000 tokens |
| Parallel requests | 1 | 2 | 4 |
| Burst protection | Standard | 500,000 | 1M |
| Queueing | – | – | ✓ |
| Weldall Desktop Client | ✓ | ✓ | ✓ |
| Transcription (Whisper) | – | ✓ | ✓ |
| Decision model | – | ✓ | ✓ |
| Embeddings | – | – | ✓ |
| Common with AI subscriptions | SeibertGPT Team |
|---|---|
| Usage limit reached, then a pause or a surcharge | Unlimited tokens, no quota |
| API costs per token on top of the subscription | Fixed price per seat, nothing else |
| More usage as a multiplier, without fixed values | Clear values: parallel requests, burst, context |
| Processing often outside the EU | Model in EU data centres, no training on your data |
Weldall Desktop Client, included in every plan
Developers plug SeibertGPT into their tools. For everyone else on the team, API access is just a key at first. The Weldall Desktop Client turns it into a workspace.
You connect your company systems with Weldall for Teams: your own Weldall instance for your team, with your systems and each person's permissions. We can run it for you, or you run it yourself. Pricing on request.
Everyone signs in with SeibertGPT. Nobody has to copy a key.
Jira, Confluence, Google Workspace and email, each with the permissions of the person asking.
Save good prompts once and reuse them across the whole team.
We only log metadata such as the model and tokens, no prompts and no responses.
Start contracts for your team right in the team area and sign them digitally.
The language model runs in data centers in the EU, the portal and API in Germany. No training on your data.
Buy as many seats as you need and pass them on when someone leaves or joins.
All seats and passes on one company invoice, payable by bank transfer if you prefer.
Extra models, NDA and DPA are already included in the plans. Discounts, term and custom plans are agreed with you in the offer.
There is no monthly quota and no top-ups. To keep everyone equally fast, Burst Protection only limits how much new context is added per minute.
3 in total, whichever plan. You can mix plans, for example one seat each of Officework, Advanced and Coding & Automation.
Yes. You can add seats at any time, charged pro rata; cancelled seats end at the end of the current period.
We set up the team and your contact person invites everyone. From then on you assign seats yourselves and hand them on when someone leaves or joins.
A reserve for new context per person that refills every minute. Advanced and Coding & Automation have a larger reserve than the default. Details in the API docs.
If someone sends more simultaneous requests than the plan allows, they are normally rejected. In Coding & Automation, requests via the API and your tools wait briefly and then run.
DeepSeek V4.1 Flash in all plans, plus Advanced: Whisper, a decision model; Coding & Automation: Whisper, a decision model, an embedding model. The team plan docs show which plan includes what.
The Weldall Desktop Client: sign in with SeibertGPT and get started, no API key needed.
One company invoice for all seats, monthly or yearly, by bank transfer on request.
Yes, depending on team size. Just ask.
Yes. When you create the NDA in your account, select “Secrecy holder under § 203 StGB, § 102 AO, § 57 StBerG or § 43a BRAO.” The NDA then includes the additional obligations for protecting professional secrets. Together with the DPA, this gives you the contractual basis.
Yes. Companies from the EU, the EEA (Iceland, Liechtenstein, Norway), Switzerland and the United Kingdom can buy online. You can also create the DPA and NDA with an address in these countries.
Yes. We sign a Data Processing Agreement (DPA) with you under Art. 28 GDPR, and your data is processed exclusively in the EU. Seibert Group GmbH is ISO/IEC 27001 certified. Learn more on the Security page.
The language model runs at our infrastructure partner Verda in data centers in Helsinki, Finland (EU), powered by renewable energy. We operate the portal and the API at Hetzner in Nuremberg and Falkenstein, Germany.
We don't store the content of API requests – neither your inputs nor the responses, only metadata for billing and limits. We store your browser chats so you can return to your history; you can delete them at any time.
No. We don't use your inputs or the responses to train AI models – and according to Verda's terms, that doesn't happen there either.
No. We run the model ourselves, and it is only accessible through SeibertGPT. Your requests don't go to the model's developer.
Yes, at no extra cost and without having to ask: right after creating your account, you set up both yourself under “Contracts” and sign them online. Both templates are available to read in advance: NDA (open template), DPA (open template). Processing takes place exclusively in the EU or the EEA. Details in the DPA.
A language model by DeepSeek that Seibert Group operates itself. You use it in the browser as a chat or via the API in your own tools and scripts.
Yes. In the chat, you can attach PDFs, Office documents and images. We convert documents to text in an isolated service; images go directly to the model.
SeibertGPT uses DeepSeek V4.1 Flash. In the meta-ranking on llm-stats.com, it is on par with Claude Opus 5 and GPT-5.6 Sol (as of October 2026) – which is usually quite sufficient for emails, summaries, research, analyses and code. Learn more on the model page.
The API is OpenAI- and Anthropic-compatible. In most tools and SDKs, you just enter the base URL and your API key – you'll find both under “API keys” after signing in. The setup guides walk you through it step by step.
The prices shown are net prices, because the offer is aimed exclusively at businesses. Companies in Germany pay 19% VAT on top; the final price is shown in small print below. Companies from other EU countries pay no German VAT with their VAT ID (reverse charge). Companies from Iceland, Liechtenstein, Norway, Switzerland and the United Kingdom receive invoices without VAT.
For companies and self-employed professionals: The offer is aimed exclusively at businesses (Unternehmer) within the meaning of § 14 BGB (German Civil Code).
No obligation and no login. Tell us how many seats you need per plan. We'll get back to you with a quote, with a discount for larger teams.
Teams are company accounts: you order as a business.
Prefer to reach out directly? Write to gpt@seibert.group.
Need more than team plans, such as your own GPU cluster? SeibertGPT for companies