Knowledge Center & News

Ideas tested in the real world.

Neur-X Insights brings together production-shaped AI benchmarks, engineering field notes, practical guides and company news. It is where we share what we learn while building reliable customer interactions for Europe.

Latest insight

Research note 01

Engineering benchmark 12 min read

Fast EU LLMs for intent detection: 19 deployments compared through a production request path

We compare 1,140 model-scenario runs across short and long conversational contexts. The results show why model size alone is a poor proxy for the accuracy, response time and cost balance a live voice or support agent needs.

Read the full benchmark
Intent30/30
Short p95338 ms
Long p95401 ms
Rows19

Fastest p95 response among perfect-accuracy deployments in each context suite.

Built to grow: future benchmarks, guides, product releases and Neur-X news will be published here in a consistent, searchable format.