Lyceum in your slides, your terminal, and more GPU capacity in Europe
Lyceum

ISSUE #004 · JULY 2026

Lyceum in your slides, your terminal, and more GPU capacity in Europe

Three things this month. You can now run Lyceum models directly inside PowerPoint, launch Claude Code preconfigured against our inference proxy, and access more B200 and B300 capacity in Europe.

Lyceum AI, Now in PowerPoint

A customer asked us to bring our models into PowerPoint. We shipped it in 4 hours and published it to everyone.

If you build decks, you can now use Lyceum AI directly in PPTX. Connect the Claude extension in PowerPoint with your Lyceum API key and use all our open-source models to design slides right where you work.

No more switching tabs, no more copy-paste. Just prompt and build.

Here is how it works:

  1. Install the Claude extension in PowerPoint.
  2. Add your Lyceum API key to connect the extension to our inference proxy.
  3. Switch between our open-source models depending on the task, drafting copy, restructuring content, or generating layout ideas, right inside your slides.

EU-hosted inference, straight into your deck.

Claude Code, Preconfigured with Lyceum

You can now run Claude Code against Lyceum’s inference proxy with zero setup friction.

Two things make it seamless:

Switching models mid-session. Change the underlying model without restarting your session. No exports needed.

A lyceum code CLI command. Launch Claude Code preconfigured against Lyceum, so you skip the manual environment setup. Still fully compatible with your existing Claude Code workflow.

Claude Code running GLM-5.2 through Lyceum's inference proxy

Lyceum in your slides, your terminal, and more GPU capacity in Europe (part 2)

That means you can run open-source models like GLM-5.2 through Lyceum’s inference proxy inside Claude Code, and switch to a different model mid-task whenever the workload changes, all on EU-hosted infrastructure.

Create your Lyceum account

More GPU Capacity Available

24-Node B200 Cluster Available in the Nordics

If your workloads fit the Nordics, this is a strong play: cheap renewable power, cooler climates, a stable regulatory environment, and now the capacity to match. Our new 24-node B200 cluster is live in Norway and Sweden, with flexible terms and multi-year commitments on the table. Nordic deployment just went from wishlist to available, and capacity like this does not sit around long.

B300 Capacity Now Available in Europe

B300 has been the tightest segment in the market all year. Most supply is locked behind long lead times or already spoken for, and everyone has been hitting the same wall. We just broke through it: B300 nodes available in Europe, no wait, ready now. While the rest of the market is still on the waitlist, our customers are already deploying. Reply to this email or book a call and we will walk you through pricing and availability, before this window closes too.

Demand Still Outpacing Supply

The demand right now is unlike anything we have seen. Reserved and on-demand capacity is selling out faster than we can bring it online, and every new cluster we announce is spoken for almost immediately. We are scaling hard, but the market is scaling harder. If you wait, someone else takes the capacity. The companies planning ahead are locking in multi-year commitments now, before the next wave of supply gets claimed. The ones who hesitate are already back on the waitlist.

I’m interested, let’s talk capacity

Not sure what GPU shape your workload needs?

Book a 30-minute validated-infrastructure assessment with our GPU solutions engineer. You walk away with a specific recommendation, not a pitch.

Book your assessment

Lyceum Technology, Berlin, Germany

Unsubscribe