Test-time training

SSI

2026-08-25 12:35:28

SSI model rumors intensify as Nvidia partnership and reported $5 billion bet draw attention

Safe Superintelligence, or SSI, the lab founded by Ilya Sutskever, is at the center of mounting speculation that it could unveil its first model as soon as this week. The discussion picked up after several investors and tech commentators posted hints online, including a16z partner Martin Casado, whose comments quickly pushed attention toward SSI because Andreessen Horowitz is among the lab’s backers. Others mentioned the possibility that the model in question could be OpenAI’s next Astra release, but the broader conversation kept returning to SSI. The report also ties the speculation to Nvidia’s July 27 announcement of a long-term strategic partnership with SSI. According to the article, Nvidia plans to expand SSI’s compute capacity by 10x over the next 12 months. It also says the investment could total $5 billion and that SSI was given exclusive access to Nvidia’s next-generation Vera Rubin systems. Nvidia said in a release that it made the commitment after gaining rare access to SSI’s tightly guarded research. At the center of the report is the claim that SSI may be building around test-time training, or TTT, an approach in which a model updates itself while processing new information instead of relying only on a larger context window. The article frames that idea as consistent with Sutskever’s public view that AI is moving beyond the pretraining era and toward systems built around continual learning.

2400

Ilya Sutskeve

2026-08-13 08:21:21

Rumor points to SSI’s first model as Ilya-linked lab explores test-time training

A new rumor circulating on X has put Safe Superintelligence, the secretive company founded by former OpenAI chief scientist Ilya Sutskever, back in focus. According to posts attributed to user San Zhi Caomei, SSI is working on a small reasoning engine built around test-time training, or TTT, a method that allows a model to keep updating part of itself while solving a problem. The same posts claim the current version is ready, that a next-generation version is being scaled up by 10x, and that an early release could come as soon as August for a limited group of users. None of those claims has been confirmed by SSI or Sutskever, and later chatter suggested a release might not happen this month after all. The rumor gained traction because it appears to line up with several public clues. Sutskever said in a November interview with Dwarkesh Patel that terms such as AGI and pretraining had pushed the field in the wrong direction, arguing that intelligence should keep learning after deployment. Nvidia’s July 27 announcement of a long-term strategic partnership with SSI added to that interest, saying the startup had spent the past two years pursuing a new research path and that its compute would rise by an order of magnitude with access to Vera Rubin systems. In that context, the latest TTT speculation has been read as a possible signal of what SSI has been building behind closed doors.

2770