Three Models for Different Needs
OpenAI has introduced the GPT-5.6 series in a limited preview, featuring three distinct models: Sol (flagship), Terra (balanced everyday model), and Luna (fast and affordable). Sol sets new state-of-the-art results on Terminal-Bench 2.1 for coding workflows and shows broad improvements in biology workflows on GeneBench v1. It also achieves strong cybersecurity performance, competing with Anthropic’s Mythos on ExploitBench while using only about one-third of the output tokens. The model introduces a new max reasoning effort and ultra mode that leverages subagents for complex work.
Enhanced Safety and Government Oversight
The launch includes OpenAI’s most robust safety stack to date, with strengthened protections for higher-risk activity, sensitive cyber requests, and repeated misuse. OpenAI dedicated over 700,000 A100-equivalent GPU hours to automated red teaming. The model uses layered safeguards including trained model-level protections, real-time generation classifiers, account-level review, and differentiated access. GPT-5.6 Sol does not cross the Cyber Critical threshold under OpenAI’s Preparedness Framework. In evaluations involving Chromium and Firefox, it identified bugs and exploitation primitives but did not autonomously produce a functional full-chain exploit.
Pricing and Availability
At the request of the U.S. government, OpenAI is starting with a limited preview for trusted partners before broader release. OpenAI stated this process should not become the long-term default as it “keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.” Pricing is set at $5/$30 per 1M tokens for Sol (input/output), $2.50/$15 for Terra, and $1/$6 for Luna. GPT-5.6 Sol will launch on Cerebras at up to 750 tokens per second in July.
Source: OpenAI
