Previewing GPT-5.6 Sol: a next-generation model
June 26, 2026 — OpenAI announced a limited preview of the GPT-5.6 model family: Sol (flagship), Terra (balanced, ~2x cheaper than GPT-5.5), and Luna (fast, lowest cost). The release follows coordination with the U.S. government after the Trump administration requested a staggered rollout due to cybersecurity concerns tied to the June 2026 AI executive order framework.
Government-Limited Preview
At the government’s request, OpenAI is starting with a limited preview for a small group of trusted partners whose participation has been shared with authorities, before broader release. OpenAI stated it does not believe this government access process should become the long-term default, but is taking the step as a path to broader availability in the coming weeks while working with the Administration on a cyber Executive Order framework.
During the preview, GPT-5.6 models are available through the API and Codex to select trusted partners. Broader ChatGPT, Codex, and API access is planned soon.
Capabilities
GPT-5.6 Sol introduces a new max reasoning effort and an ultra mode leveraging subagents for complex work. Highlights include:
- Coding: State of the art on Terminal-Bench 2.1 (command-line agent workflows)
- Biology: Stronger results on GeneBench v1 vs GPT-5.5 with fewer tokens
- Cybersecurity: Competitive with Mythos Preview on ExploitBench² using ~1/3 output tokens; improvements on ExploitGym across Sol, Terra, and Luna
OpenAI reports GPT-5.6 Sol achieved 96.7% on internal cyberattack challenge tests. All three GPT-5.6 models are classified as High capability in cybersecurity and biological-and-chemical risk under the Preparedness Framework.
Safeguards
OpenAI deployed its most robust safeguard stack to date: model-level refusals, real-time cyber/biology misuse classifiers, account-level review, differentiated access, and 700,000+ A100-equivalent GPU hours of automated red-teaming plus third-party human red teaming. Safeguards may occasionally block legitimate dual-use work during the preview.
Pricing and Availability
Per 1M tokens:
- Sol: 30 output
- Terra: 15 output
- Luna: 6 output
GPT-5.6 introduces improved prompt caching (explicit cache breakpoints, 30-minute minimum cache life). GPT-5.6 Sol on Cerebras at up to 750 tokens/second is planned for July with initial limited access.