Eval Console
The content checks in Dvar's Eval Console: traffic, answer key, rubric, compare, fine-tune
The product a team would use every day, on the same test scenes as the bench. Read-only in this demo.
Internal benchmark on synthetic data · run run-202610230802 · 2026-10-23 · 7 model runs, same instruction per task
What to look at: Every scene the content check read, with today's model's verdict, its topic (the show's genre), language, reply time and cost. Click a row to open its trace.
The Eval Console is built for a laptop screen. On a phone, scroll inside it; the Bench tab has every result.