OpenAI has announced that its newly released GPT-5.6 Sol Ultra has successfully proved the 50-year-old Cycle Double Cover Conjecture in under one hour, utilizing 64 subagents for the task. This accomplishment highlights the model’s capability to parallelize complex work, as it directed multiple subagents toward different approaches while conducting repeated audits. The release of GPT-5.6 Sol Ultra is expected to enhance user engagement and innovation in problem-solving.

OpenAI: OpenAI develops and releases advanced AI models focused on reasoning and problem-solving capabilities. The company announced the general availability of GPT-5.6 Sol Ultra and highlighted its use of multiple subagents to tackle complex tasks. This positions OpenAI at the forefront of exploring parallelized AI approaches for scientific and mathematical challenges.
Eknight: Eknight is the OpenAI representative who publicly announced the general availability of GPT-5.6 Sol Ultra and shared details of its performance on a major conjecture. The announcement emphasized the model’s subagent architecture and included the relevant prompt and proof for community review. Eknight highlighted excitement about applications of the new capabilities.
GPT-5.6 Sol Ultra: GPT-5.6 Sol Ultra is OpenAI’s model designed to handle intricate problems by coordinating multiple subagents that explore competing strategies and perform iterative reviews. It was recently made generally available and demonstrated the ability to generate proofs for longstanding mathematical conjectures through structured prompting. The model emphasizes rejection of incomplete arguments to ensure rigorous outcomes.

`json
{
“Model Release”: “OpenAI has made GPT-5.6 Sol Ultra generally available to users.”,
“AI Architecture”: “GPT-5.6 Sol Ultra supports parallel processing by utilizing multiple subagents to approach complex tasks.”
}
`