On October 10th, according to The Insider of BusinessGoogleA new generation will be officially released to the public Gemini 4"Argon”ModelI don't know. At the same time, its in-house staff are already testing a later iterative version with more performance, which is expected to further narrow the technology gap with the competition。
According to internal files and screenshots obtained by the media, Google staff have recently been testing a new version of Gemini 4 codenamed “Carbon”. The model is likely to be an update of the new checkpoint (Checkpoint) under the Argon architecture and has recently been accessed on the Google internal code platform Jetski。
One employee revealed that Carbon's performance on code capabilities was “belike Opus 5.5” -- the strongest model under the Anthropic flag for a long-term mission of complex code intelligence. At the same time, however, the employee indicated that this conclusion still needed to be validated in greater depth。
Based on a combination of multiple staff feedback and internal communication records, earlier versions of Argon generally yielded good results in internal testing. It was noted, however, that the earlier version was still ferocious in dealing with some programming tasks, which was roughly equivalent to Claude Opus 5, released earlier by Anthropic。
An internal document indicates that multiple Gemini 4 derivative models codenamed Argon, Barium and Carbon have been tested within Google。
It is worth noting thatGoogle's internal code does not always relate to the final public nameI don't know. For example, it has been documented that a model previously selected for external publication under the name “Argon” is internally coded as “Barium-B”. Thus, it remains to be determined whether the current Carbon will be introduced as a follow-up fine-tuning version of Argon, or as a new independent brand under the Gemini 4 family。
Google refused to comment on this。