OpenAI has begun a limited preview of GPT-5.6, a new model family spanning three tiers: Sol, the flagship; Terra, a balanced everyday model; and Luna, a fast and affordable option. Announced on June 26, 2026, the release pairs OpenAI's most capable model yet with what the company calls its most robust safety stack to date.
One family, three durable tiers
Under the new naming system, the number marks the generation while Sol, Terra and Luna identify capability tiers that can each advance on their own cadence. Terra matches GPT-5.5-class performance at roughly half the price, and Luna brings strong capability at OpenAI's lowest cost. GPT-5.6 also adds a new max reasoning effort and an ultra mode that spins up subagents to parallelize complex, multi-step work.
State-of-the-art coding, biology and cyber
OpenAI says Sol sets a new state of the art on the Terminal-Bench 2.1 coding benchmark, improves long-horizon genomics analysis on GeneBench, and is its most capable model yet for cybersecurity tasks such as vulnerability research. The company stresses Sol is better at helping defenders find and fix vulnerabilities than at carrying out end-to-end attacks, and says it does not cross the Cyber Critical threshold of its Preparedness Framework. The launch follows OpenAI's broader push to own its stack, from its first custom inference chip with Broadcom to the recent general availability of GPT-5.5 on Amazon Bedrock.
Phased, government-coordinated rollout
During the preview, the models are available through the API and Codex to a small group of trusted partners whose participation was shared with the U.S. government ahead of launch. OpenAI says it does not believe such government access should become the long-term default and plans general availability "in the coming weeks," with a Cerebras deployment of Sol at up to 750 tokens per second slated for July. Pricing per million tokens runs $5/$30 for Sol, $2.50/$15 for Terra and $1/$6 for Luna. The careful rollout echoes the heightened scrutiny across the sector as labs race toward public markets, including Anthropic's confidential IPO filing.
Reporting based on coverage from OpenAI, VentureBeat and Axios.