Former OpenAI Chief Scientist Ilya’s SSI Unveils First Model

AI has achieved another major milestone. Following his departure from OpenAI, former chief scientist Ilya Sutskever established Safe Superintelligence Inc. (SSI), which has recently unveiled details regarding its inaugural model. Reports from overseas social media suggest the team is developing a compact reasoning engine leveraging TTT (Training While Testing), a crucial advancement in their pursuit of safe superintelligence.
This novel architecture focuses on enabling models to "learn how to learn." In contrast to conventional large language models that keep parameters static during inference, this engine dynamically updates specific weights while addressing real-world tasks. This approach reflects Sutskever’s belief that true superintelligence requires continuous evolution post-deployment, rather than relying solely on static training. Despite historical security hurdles associated with TTT, recent reports indicate these technical challenges have been overcome.
Regarding the timeline, this initial version may be accessible to a curated group of invited testers as early as August. Furthermore, the subsequent model iteration is projected to scale up tenfold. Although SSI has maintained strict secrecy, the disclosed information aligns with previous industry indicators concerning their technical roadmap, progress, and partnerships with tech giants like NVIDIA. With this innovative concept materializing, the AI sector may soon witness a disruptive product capable of "adapting tactics in real time."
Related article
Anthropic Launches Claude Haiku 5.5, Cutting Small Model Costs by 75% and Letting Users Adjust Inference Intensity
On October 8th, Anthropic introduced its most efficient and capable model to date: Claude Haiku 5.5. Positioned as the fastest and most capable small model in Anthropic’s lineup, it is engineered for high-frequency, cost-sensitive applications such a
Claude Unveils First Full Ultraviolet Sky Map, Charting 119 Million Stars
With the powerful assistance of Claude, Anthropic’s large language model, astrophysicists have successfully generated the first complete ultraviolet all-sky map in history. This vast star chart, featuring 119 million stars, was largely completed by C
How AWS and NVIDIA Are Powering Agentic and Physical AI
NVIDIA CEO Jensen Huang (Credit: Tomohiro Ohsumi/Getty Images)Jensen Huang’s NVIDIA and AWS scale their global partnership, deploying 2M GPUs to power next-generation agentic and physical AI enterprise infrastructureAWS and NVIDIA have announced a ma
Related Special Topic Recommendations
Comments (0)
0/500

AI has achieved another major milestone. Following his departure from OpenAI, former chief scientist Ilya Sutskever established Safe Superintelligence Inc. (SSI), which has recently unveiled details regarding its inaugural model. Reports from overseas social media suggest the team is developing a compact reasoning engine leveraging TTT (Training While Testing), a crucial advancement in their pursuit of safe superintelligence.
This novel architecture focuses on enabling models to "learn how to learn." In contrast to conventional large language models that keep parameters static during inference, this engine dynamically updates specific weights while addressing real-world tasks. This approach reflects Sutskever’s belief that true superintelligence requires continuous evolution post-deployment, rather than relying solely on static training. Despite historical security hurdles associated with TTT, recent reports indicate these technical challenges have been overcome.
Regarding the timeline, this initial version may be accessible to a curated group of invited testers as early as August. Furthermore, the subsequent model iteration is projected to scale up tenfold. Although SSI has maintained strict secrecy, the disclosed information aligns with previous industry indicators concerning their technical roadmap, progress, and partnerships with tech giants like NVIDIA. With this innovative concept materializing, the AI sector may soon witness a disruptive product capable of "adapting tactics in real time."
Anthropic Launches Claude Haiku 5.5, Cutting Small Model Costs by 75% and Letting Users Adjust Inference Intensity
On October 8th, Anthropic introduced its most efficient and capable model to date: Claude Haiku 5.5. Positioned as the fastest and most capable small model in Anthropic’s lineup, it is engineered for high-frequency, cost-sensitive applications such a
Claude Unveils First Full Ultraviolet Sky Map, Charting 119 Million Stars
With the powerful assistance of Claude, Anthropic’s large language model, astrophysicists have successfully generated the first complete ultraviolet all-sky map in history. This vast star chart, featuring 119 million stars, was largely completed by C
How AWS and NVIDIA Are Powering Agentic and Physical AI
NVIDIA CEO Jensen Huang (Credit: Tomohiro Ohsumi/Getty Images)Jensen Huang’s NVIDIA and AWS scale their global partnership, deploying 2M GPUs to power next-generation agentic and physical AI enterprise infrastructureAWS and NVIDIA have announced a ma





Home






