The tools they already use
- →ChatGPT Desktop and Codex
- →Word, Excel and PowerPoint
- →Every OpenAI-compatible client
- →Nothing to relearn
← Part of Walma AI HubAI Hub · OpenAI
ChatGPT in Europe. Implemented by Walma.
GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna for your whole organisation, in an environment you own inside the EU, with every euro accounted for per team, person and project.
Azure OpenAI in an EU regionChatGPT Desktop, Codex and OfficeCost per team in real time
The model catalogue
Every OpenAI model, one key
The whole GPT family behind the same gateway. You pick a default model per team, set a budget and switch when OpenAI ships the next one.
| Model | Best for | Context | Price per 1M tokens (in / out) | EU region |
|---|---|---|---|---|
| GPT-6 Astragpt-6-astra | The hardest work: agent flows, code, research and long documents. | 1.05M | $10 / $50 | Global until EU zone |
| GPT-6.1 Solgpt-6.1-sol | Close to Astra at a fifth of the price: analysis, reports, legal and code. | 1.05M | $2 / $10 | EU data zone |
| GPT-6 Lunagpt-6-luna | Volume: classification, extraction and summaries at scale. | 1.05M | $0.10 / $0.50 | EU data zone |
More models in the catalogue
CodexGPT-6 Astra · GPT-6.1 Sol · GPT-6 Luna
Coding agents in the terminal and IDE, on GPT-6 Astra, GPT-6.1 Sol or GPT-6 Luna.
OpenAI DaybreakGPT-5.6 Cyber · Daybreak Red · Daybreak Blue
Authorised security testing and vulnerability research.
GPT-Image-2.5Sunburst · Flare
Image generation and editing. Sunburst for quality, Flare for speed.
GPT-Live 1GPT-Realtime-2.1 · Realtime-2.1 Mini · Translate
Voice assistants and real-time conversations with tool calls.
GPT-TranscribeGPT-Live-Transcribe · Realtime-Whisper
Transcription of meetings and calls, in real time or from file.
EU region means the Azure OpenAI EU Data Zone, which includes Sweden Central, as of October 2026. GPT-6 Astra currently runs via global routing, and you choose per team whether it is allowed. Prices are OpenAI list prices in USD.
ChatGPT Enterprise or Walma?
The same models. Your environment.
ChatGPT Enterprise stores data in Europe, but inference happens in the US by default and the licence is paid per seat. With Walma, GPT runs via Azure OpenAI in an EU region in your own tenant, and you pay for what is actually used.
| ChatGPT Enterprise | GPT via Walma | |
|---|---|---|
| Data storage | EU (data residency) | EU, in your own tenant |
| Inference | US by default, EU for selected customers | EU region (Sweden Central or EU Data Zone) |
| Pricing model | Licence per seat | Actual usage, budget per team |
| Other models | OpenAI only | Claude, Mistral, Llama and more behind the same key |
| Logging and cost | Admin panel per workspace | Per team, person and project, in your currency |
| Policy | Settings per workspace | Allowlists, caps and rules in the gateway |
The full package
GPT where the work already happens
No new window. Your teams use ChatGPT Desktop, Codex and the Office add-ins as usual, but through your environment.
Connected to your systems
- →Documents in SharePoint and OneDrive
- →Teams and Outlook
- →Your internal systems via API
- →Your own knowledge in the answers
Governed from day one
- →Your existing sign-in (SSO and SCIM)
- →Model allowlist per team
- →Guardrails against data leakage
- →Everyone onboarded in minutes
Value in numbers
- →Cost per team, person and project
- →Budgets and quotas in real time
- →Automatic downgrade Astra, Sol, Luna
- →Numbers you can show the board
The problem
AI use is growing. Control isn't keeping up.
Developers reach for whatever model works. Finance sees the bill a month later. Security finds out when the data has already left the building.
Shadow AI
Personal API keys, unapproved tools, no attribution. You can't govern what you can't see, and right now you can't see any of it.
Surprise invoices
Usage-based pricing with no budgets means the cost only becomes real when it's already too late to do anything about it.
Data leaving the EU
Every prompt through a US endpoint is a GDPR question you don't want to answer in an audit, and most teams can't even tell where it went.
How you get started
From decision to the whole organisation
Four weeks from decision to rollout. We set up Azure OpenAI in your tenant, connect your systems and drive adoption together with you.
- 01
We map the current state
Week 1Who already uses ChatGPT, with which accounts and for what. The plan starts from your reality, not a template.
- 02
Your environment is set up in the EU
Week 1 to 2Azure OpenAI in the region you choose, connected to the sign-in your people already have. The models are in place the same day.
- 03
Rules and budget
Week 2Allowlists, budgets and guardrails per team, set together with you.
- 04
Rollout and training
Week 3 to 4Your teams get GPT in the tools they already use. We train and make sure usage takes off.
- 05
The numbers are followed up
OngoingWalma Studio shows usage, cost and value per model. We review the numbers with you continuously and adjust where needed.
Cost calculator
What is AI actually costing you?
Drag the sliders for a back-of-the-envelope estimate, then see what governance claws back.
Estimated AI spend
1 940 €/ mo
With Walma: caching, auto-downgrade, no waste1 319 €
You could save
≈ 621 € / month
Rough estimate on typical token usage. Book a demo for your real numbers.
Cumulative spend over 12 months
Without WalmaWith Walma
FAQ
Questions about ChatGPT via Walma
Can we run GPT-6 Astra in the EU?+
GPT-6 Astra was released on 3 September 2026 and is, at the time of writing, available on Azure with global routing only, not in the EU data zone. GPT-6.1 Sol and GPT-6 Luna run in the EU data zone, which includes Sweden Central, today. In Walma you choose per team whether Astra is allowed, and the log shows exactly which calls left the EU.
What is the difference between Astra, Sol and Luna?+
Three tiers of the GPT-6 generation. Astra is the most capable, Sol (version 6.1) comes close to Astra at a fifth of the price, and Luna is the fastest and cheapest for volume. All three have a context window of just over a million tokens. The older GPT-5.6 Sol, Terra and Luna remain in the API but are no longer OpenAI's recommended models.
Does Walma replace ChatGPT Enterprise?+
For most organisations, yes. You get the same models in your own EU environment, pay for actual usage instead of per seat, and can add Claude and other models behind the same key. If you have an Enterprise contract you want to keep, it can be plugged in.
Do ChatGPT Desktop and Codex work?+
Yes. ChatGPT Desktop, Codex and every OpenAI-compatible client are pointed at the gateway with a base URL and key. Signed installers for Windows and macOS make it a matter of minutes.
What does it cost?+
You pay OpenAI list prices for usage, with no markup, plus a flat per-seat fee for the Walma governance layer. Most save more on downgrades and budgets than the platform costs.
AI Hub · OpenAI
Thirty minutes. Then you know where you stand.
We go through how your teams use ChatGPT today and calculate what the same work yields once it gets an EU environment you own.