OpenAI is opening early access to GPT-5.6: three models, three levels of capabilities, and enhanced security measures.
OpenAI has taken a significant step in the development of its language models by introducing the GPT-5.6 family, consisting of three specialized versions: Sol, Terra, and Luna. Currently, access via the API and Codex is only open to a limited circle of trusted partners, and this launch was coordinated with the US government, which requested a preliminary preview phase.
Range of Capabilities: From Flagship to Budget Solution
The flagship model Sol is positioned as the most powerful in the company's history. It is equipped with a max reasoning mode, allowing more time for deep processing of complex tasks, as well as an ultra mode that uses sub-agents to accelerate complex operations. Terra, in turn, is described as a "workhorse" for everyday tasks. According to OpenAI estimates, it demonstrates performance comparable to GPT-5.5 but costs half as much. Luna is the fastest and most economical option, designed for mass use.
Performance Records and Cybersecurity
Sol has already set a new record in the Terminal-Bench 2.1 benchmark for command-line tasks. On GeneBench v1, the model showed results higher than GPT-5.5 with lower token consumption. In ExploitBench tests, Sol proved competitive with Mythos Preview while using approximately one-third of the output tokens. Notably, all three models show improved results on ExploitGym as reasoning depth increases.
In terms of security, OpenAI claims that the GPT-5.6 family has received the strongest protection stack. Sol does not cross the Cyber Critical threshold within the Preparedness Framework. In tests with Chromium and Firefox, the model successfully found bugs but could not autonomously create a fully functional exploit. Over 700,000 GPU-hours (in A100 equivalent) were used for automated red teaming to verify protection.
Pricing and Future Plans
The cost per 1 million tokens is: for Sol — $5 for input and $30 for output; for Terra — $2.50 and $15; for Luna — $1 and $6. In July, OpenAI plans to launch Sol on the Cerebras platform with speeds of up to 750 tokens per second, but access at this stage will also be limited.
My comment as an analyst: The division into three clear tiers — Sol, Terra, Luna — is a smart strategic move that allows OpenAI to cover different market segments, from enthusiasts and researchers to large corporations. However, the fact that the launch is taking place under the control of the US government highlights regulators' growing concerns about the safety of powerful AI models. In the coming weeks, we will see how the community receives these new products, but it is already clear that OpenAI is betting on controlled rather than mass adoption.