OpenAI’s next GPT model is coming after an unusual pause.
GPT-5.6 is expected to roll out this week after US national security concerns reportedly delayed broader access last month. The model family includes GPT-5.6 Sol, OpenAI’s most capable version, along with Terra and Luna, lower-cost models built for wider use.
For businesses, developers, and security teams, the launch brings stronger coding, cybersecurity, and agentic workflows while adding a new layer of uncertainty around government review, export controls, and enterprise AI governance.
GPT-5.6 heads for a wider launch
According to Reuters, GPT-5.6 Sol is OpenAI’s most advanced model. OpenAI previewed improved agentic capabilities in coding, biology, and cybersecurity in late June and said GPT-5.6 Sol will be competitive with Anthropic’s Mythos Preview on the ExploitBench cybersecurity benchmark.
CNBC, citing Axios, reported that the US Department of Commerce cleared OpenAI to proceed with a broad GPT-5.6 release after additional testing and meetings between the company and government officials. OpenAI had limited access to a small group of vetted partners after US government requests delayed a broader rollout.
The AI firm previously said it started a limited preview of the GPT-5.6 series, with Sol as its flagship model, Terra as a balanced model for everyday work, and Luna as its faster, lower-cost option. The company also announced plans to make the models generally available in the coming weeks.
AI model launches are becoming policy events
The delayed rollout showed how frontier AI releases are moving closer to government review. US officials increased scrutiny of advanced AI models due to concerns about cyber operations, military planning, or intelligence work.
Reuters said that President Donald Trump signed an executive order creating a voluntary framework for developers to provide “covered frontier models” to the US government for up to 30 days before release to trusted partners.
For enterprise AI buyers, the delay may change how future model rollouts are planned. Access to newer models may depend on vendor timelines, security testing, government review, export controls, and model-specific safeguards.
Companies building AI features into customer support, coding workflows, and security operations may need more flexible model governance plans if frontier releases become phased or restricted.
Stronger models bring tighter controls
OpenAI highlighted that GPT-5.6 Sol includes its most robust safety stack to date, with protections for higher-risk activity, sensitive cyber requests, and repeated misuse. The company also said the model family was designed to preserve legitimate uses such as code review, vulnerability research, patch development, debugging, security education, and defensive testing.
A stronger model could help developers test software, analyze code, write patches, and investigate vulnerabilities faster. It could also create more review work for security leaders who must decide which users get access, what data can be entered, and which workflows need logging or approval.
The Anthropic comparison shows the same tension.
Reuters reported that Anthropic disabled Mythos 5 and Fable 5 after a US export control order, then restored access to some models after adding safeguards. Anthropic also warned that it was “probably impossible” to make any AI model fully robust against jailbreaks.
Before teams move sensitive workflows to GPT-5.6, they should review model access, approved use cases, audit logs, data controls, and red-team testing. The upgrade may bring better tools, but it also gives enterprises another reason to tighten AI governance before adoption spreads across departments.
Learn more about a lifetime AI tools deal that brings GPT-4o, Claude, Gemini, and more into one platform for $55.