< Back to announcements

September 2, 2025

Tomato v0.1 Test Release

We officially released Tomato v0.1, a test release built on top of an upgraded training process based on chatgpt-oss-120b, marking Tomato's first public step as an independent product direction.

What we released

Today, we officially released Tomato v0.1.

This is the first public test version of Tomato. Built through an upgraded training process based on chatgpt-oss-120b, this release marks Tomato's first step from internal exploration toward becoming an independent product.

At the v0.1 stage, our focus is not on scale. Instead, we are concentrating on a few of the most fundamental and most important questions:

  • Whether model output quality meets expectations
  • Whether the core inference pipeline is stable enough
  • Whether the product experience is strong enough to justify continued iteration
  • Whether we can collect meaningful feedback from real usage scenarios

Why this matters

Every long-term product needs a real starting point.
The significance of Tomato v0.1 is not that it is already complete, but that it allows us to begin iterating based on real usage instead of remaining at the concept stage.

What comes next

Over the next few weeks, we will continue shipping updates based on early testing feedback, with a focus on:

  • Inference stability
  • Response quality
  • Core product capabilities
  • The engineering foundation required for future commercialization and service delivery

This is Tomato's first step, and only the beginning.