Gemini 4 Argon spotted in A/B testing on Antigravity

A new tracking report says an entry labeled Gemini 4 Argon was spotted running through A/B testing on the Antigravity platform. It is a trace from the pre-launch testing stage rather than an official Google announcement.

The report is thin by design. Its core claim is narrow: a model referred to as Gemini 4, alongside the codename Argon, was observed in an A/B test environment tied to Antigravity. Google has published no official statement about such a model, and no capabilities, parameters, or schedule are attached to the sighting.

A/B testing is standard practice for large AI developers. A new build is exposed to a small slice of users so the team can watch real usage, catch regressions, and decide whether to widen the rollout. Seeing a version at that stage suggests something runnable exists internally, but it does not mean a public release is imminent.

TestingCatalog and similar trackers work by watching interface changes, feature flags, and gradual rollouts across vendor products. Their upside is an early paper trail that often precedes official announcements; their limit is that a codename reveals little about the finished product.

For anyone following Google's model roadmap, the interesting signal is the testing cadence itself. If a Gemini 4 build is already in an A/B environment, the next generation is moving through internal stages even while the public picture stays deliberately quiet.

It is worth keeping the finding in proportion. Until Google confirms anything, the naming, the capability boundaries, and any release window remain open questions. The things to watch are an official acknowledgment and whether the build graduates from a limited test into broader availability.