The US government has imposed restrictions on OpenAI’s latest and most advanced AI models, the GPT-5.6 family, which includes Sol, Terra, and Luna. These models, known for their varying capabilities, have been limited to a select group of trusted partners following a government request. This decision, announced by OpenAI, has sparked discussions about governmental control over the release of advanced AI technologies. Dean Ball, a former White House AI adviser, highlighted that a new executive order mandates a review process for advanced models 30 days prior to their release, potentially delaying launches and impacting competition in the global AI landscape. OpenAI has expressed its criticism of this arrangement, hoping for a broader release of the models in the near future.
Simultaneously, Anthropic has announced the partial redeployment of its Claude Mythos 5 model, focusing on cybersecurity for US organizations managing critical infrastructure. The company is collaborating with the US government to restore public access to its Mythos 5 and Fable 5 models.
The GPT-5.6 Sol model has excelled in coding performance tests, integrating advanced reasoning modes and safety guardrails. In terms of pricing, Sol is positioned as a premium option, with Terra and Luna offering more cost-effective alternatives. OpenAI has also optimized prompt caching to reduce costs for repeated use.
