OpenAI unveils GPT-5.6 family of models with three tiers: Sol, Terra, Luna

OpenAI unveiled its newest family of AI models, GPT-5.6, on Thursday, consisting of three variants: Sol, positioned as the workhorse; Terra, an intermediate option; and Luna, the budget-friendly choice. The company promises enhanced capabilities across enterprise work, coding, and scientific research.

CEO Sam Altman has stated that the new models are orders of magnitude more efficient and cost-effective than previous versions, with Sol being 54% more token efficient for AI coding tasks. OpenAI claims GPT-5.6 is its strongest cybersecurity model yet, achieving frontier performance with significantly fewer tokens. The Trump administration previously sought to restrict its rollout due to fears of misuse.

The model supports defensive activities such as threat modeling, code review and patching, and blue teaming—simulating attacks on one's own systems to find vulnerabilities. OpenAI also launched ChatGPT Work, a workplace companion for enterprise teams available on desktop, web, and mobile, assisting with clerical tasks like drafting documents, spreadsheets, and presentations.

The release follows similar announcements from SpaceXAI and Meta this week. However, GPT-5.6 appears aimed primarily at rival Anthropic. Citing the Artificial Analysis Coding Agent Index, OpenAI claims its models outperform Anthropic's. Specifically, Sol, described as OpenAI's best coding model yet, scores 80 on the index, 2.8 points above Anthropic's Fable 5, while using less than half the output tokens, taking less than half the time, and costing about one-third less. Terra performs just above Fable 5, while Luna outperforms Opus 4.8.

Pricing per million tokens is: Sol – $5 input / $30 output, Terra – $2.50 input / $15 output, Luna – $1 input / $6 output. The models are available across ChatGPT, Codex, and the OpenAI API.