OpenAI Releases New Flagship Model 「GPT-5.6 Sol」

OpenAI has released its new flagship model 「GPT-5.6 Sol」. The model reportedly demonstrates superior performance compared to Anthropic's 「Claude Mythos 5」 on coding benchmarks. However, the deployment has been provided in a restricted form due to U.S. government regulations, and OpenAI has expressed dissatisfaction with this situation.

Regarding AI development, the U.S. government has increasingly imposed conditions on the broad public release of certain high-performance models for reasons of national security and public interest. Government access restrictions apply to the deployment of GPT-5.6 Sol as well. OpenAI has characterized these rules as "unsustainable," suggesting that tensions have emerged between the company and regulatory authorities.

While the specific details of the restrictions remain unclear at this point, the deployment scope and available user base are effectively narrowed. OpenAI's official expression of concern using the term "unsustainable" demonstrates that it views this not merely as a technical issue, but as a matter directly affecting its business strategy.

OpenAI and Anthropic continue to compete on the performance of large language models (LLMs), with the GPT and Claude series as their respective cornerstones. Coding benchmarks serve as widely-referenced indicators of capability for software development support applications, and the current results directly relate to competitive positioning in the enterprise market. However, it is important to note that superior benchmark results do not necessarily translate directly to practical advantages, and ultimately, actual user experience, pricing, and availability conditions become the decisive factors in selection.

What this deployment reveals is the reality that the public release of cutting-edge AI models is no longer determined purely by technical and commercial considerations. As government involvement strengthens, AI companies are entering a new phase in which they must balance regulatory relationships while conducting business. The approach that OpenAI takes in response to these restrictions in the future could serve as a precedent affecting other model deployments, making it a matter of significant attention.