New Delhi (Reuters) – AI pioneer OpenAI has signaled that its most sophisticated model, GPT‑5.6, will be publicly released on Thursday, ending a brief hold imposed last month after U.S. officials raised national‑security alarms over the model’s potential misuse. The company said the launch will include the flagship Sol variant together with lower‑cost Terra and Luna versions.

Regulatory backdrop and geopolitical pressure

The United States and China are locked in a race to develop cutting‑edge AI systems that could accelerate cyber‑attacks, biotech breakthroughs, and complex code generation. Washington has tightened scrutiny of advanced model releases, fearing that adversarial states could weaponise the technology. Parallelly, Chinese authorities have convened with domestic tech giants to contemplate restricting foreign AI access, especially for models not yet released.

Competitor reactions

Anthropic, a direct rival, abruptly disabled its flagship models Mythos 5 and Fable 5 after a June 12 export‑control order cited national‑security risks. Those restrictions were lifted after Anthropic installed additional safeguards, yet some models remain limited to “trusted” U.S. entities. Meanwhile, Elon Musk’s SpaceXAI rolled out its own leading model, Grok 4.5, further intensifying market competition.

OpenAI’s market strategy

OpenAI disclosed plans to lower subscription and API pricing to stay competitive in a rapidly expanding AI market. In a late‑June preview, the firm highlighted GPT‑5.6 Sol’s enhanced agentic abilities in coding, biology, and cybersecurity, noting that it outperformed Anthropic’s Mythos preview on the ExploitBench cybersecurity benchmark.

Future implications and challenges

President Donald Trump’s recent executive order establishes a voluntary framework requiring AI developers to provide “covered frontier models” to the U.S. government for up to 30 days before broader release. While this aims to curb misuse, Anthropic warned that achieving full robustness against jailbreaks may be “probably impossible.” As GPT‑5.6 enters the public sphere, it will test the balance between innovation, security oversight, and international AI power dynamics.