A new model quietly joined Arena. Does its tone match? Don't judge it from one or two rounds.

The codename does not show which company made the model, and Arena is a blind test, so judging its style depends on what you ask it.

2026-10-29 · 1 min read · source @leo114119

A new model showed up on Arena. Its test codename is fe-pichu. The codename does not show which company built it. Someone who read its answers said the tone sounded strong and full. It also seemed a bit stiff and formal, with a sense of weight.

Arena is a blind test. You do not know which company's model is behind each answer. You only see which of two answers is better. This makes the test cleaner, because no brand name shapes your judgment. What you see is mainly the answer itself.

Whether the tone matches depends on what you ask. Chat and coding may bring out very different styles. One or two rounds is too few. Drawing a conclusion now would be too early.

Anonymous tests like this are a window for seeing new models early. Write down your first impression. Then ask several kinds of questions to check it. That is far more reliable than taking someone else's word for it.