Essential
Start smart — access powerful LLMs with simplicity and security.
Pricing
Two products, two ways to get started. Pick a LLM Serving plan that fits your workload, or talk to us about SmartChat to scope the right setup for your team. Custom and on-premises hosting available on request.LLM Serving
Plans that scale with your use case
Each tier adds capability, not just capacity. Start with Essential and move up as your work gets more ambitious.MOST POPULAR
Professional
Scale your AI — advanced features for growing teams and projects.
Agentic
Automate intelligently — build and deploy AI agents for complex workflows.
- *
- Input/Output Token Price:Example pricing for T-Cloud-hosted Large Language Models. Prompt tokens served from the prefix cache are billed at 15% of the input price.
- **
- Monthly minimum:You're billed the minimum or your actual usage — whichever is higher. Usage counts toward the minimum, not on top.
- ***
- Requests per minute (RPM):Example RPM for a representative T-Cloud model — exact limits vary by model
SmartChat
Built around your use case
Two pricing options: usage-based or seat-based, plus special conditions for large-scale rollouts. Tell us what you're building and we'll put together the right plan.Frequently Asked Questions
Find clarity on our plans, pricing, and capabilities — so you can focus on building.No. Each model has the same input and output token rates on every plan. Plans differ in rate limits and which models you can call — not in price per token.
Higher rate limits and broader model access. Essential includes the Standard model catalogue at up to 300 requests per minute; Professional and Agentic add the Premium models at up to 600 and 1,000 requests per minute respectively. Token prices are the same on every plan. Published RPM values are best-effort ceilings, not part of the SLA.
Yes. Every model has separate rates per million tokens — one for input (your prompt), one for output (the response). Output is usually higher. GPT-OSS 120B, for example, runs €0.20 for input and €0.65 for output. Both rates are listed on every model's row.
Yes. There's no minimum contract term: cancel with one month's notice to the end of the month. Plan changes are handled on request. Enterprise agreements have custom terms.
Yes. Sign up for a free API key: 5M tokens over 3 months, no billing setup, no credit card.
No. Prompt tokens served from the prefix cache are billed at 15% of the normal input token price. This applies to T-Cloud-hosted models today; cached pricing for models on external providers follows later. The cache is on by default and you can switch it off.
Yes. Preview models — currently GLM 5.3 Flash and Qwen 3.8 27B — are free to use on any active plan and run on T-Cloud in Germany. They sit outside the SLA: Telekom can limit rate limits, token volume or availability, or withdraw them, at any time.
Have a question? Talk to our team.
Tell us about your use case, or dive into the technical docs to get started.Opens an email to ai@t-systems.com