Services
AI models
We adapt an open model to your documents through fine-tuning. You use it locally on your hardware.
Classes
Suited for
Model M
Knowledge queries, document work, draft texts and answers based on your documents.
- Base model
- Qwen3.5-9B
- Size
- 9 bn parameters
Suited for
Model L
Multi-step domain logic, AI agents and demanding tasks with many dependencies.
- Base model
- Qwen3.8-27B
- Size
- 27 bn parameters
Suited for
Custom
Several models, groups of companies or special requirements. We set scope and fixed price after the first conversation.
Further development
Retuning and upgrade
Retuning adapts an existing model on new company data, with an evaluation against the previous version and a test report. The upgrade from M to L builds on model L, with the existing and new data.
What every model includes
The basis is open models under the Apache 2.0 licence. We set the type and scope of the data in the quote.
Data preparation and filtering
Generation of the training data
Fine-tuning
Evaluation with test report
Delivery for local use
Which class fits your project, we clarify in the free first conversation. After that you receive a quote at a fixed price.