Model updates for evroc Think - May 2026
We recently completed a model refresh across evroc Think Shared and Dedicated Models. A key shift with this update is stronger support for agentic workflows, where models need to reason, use tools and solve tasks across multiple steps.
The latest update introduces two new models selected for their practical strengths across language, vision and coding workloads:
- Gemma 4 (26B) from Google
- Qwen 3.6 (35B) from Qwen Note: Qwen 3.6 will replace its respective, older versions.
Gemma 4 and Qwen 3.6 offer cost-efficient, small models with strong performance across a broad range of use cases.
As part of the same upgrade cycle, the following models have been removed from evroc Think Shared Models:
- Phi-4 (15B) from Microsoft
- Devstral Small 2 (24B) from Mistral
- Magistral Small 1.2 (24B) from Mistral These models remain available through evroc Think Dedicated Models. If you are on evroc Think Shared Models and using any of these models, consider migrating to one of the smaller models we just launched support for.
We expect to refresh our Mistral model range in the near future.
Written by
Mikael manages evroc’s product offerings across AI, Storage, and IAM. He brings deep experience from open source, technical writing, and launching products that make complex technology easier to use.
The European Cloud
