OpenAI unveiled its latest family of models, GPT-5.6, on Thursday, introducing three versions: Sol, Terra, and Luna. These models are designed to enhance capabilities in enterprise work, coding, and scientific research.
CEO Sam Altman stated that Sol offers a 54% increase in token efficiency for AI coding tasks compared to previous versions. OpenAI claims GPT-5.6 is its strongest cybersecurity model, achieving high performance with fewer tokens.
The model’s cybersecurity features support defensive activities such as threat modeling, code review, and blue teaming, which involves simulating attacks to identify system vulnerabilities. The rollout of GPT-5.6 faced scrutiny, with the Trump administration previously seeking to restrict it due to potential misuse concerns.
OpenAI also announced a new tool, ChatGPT Work, which aims to assist enterprise teams with clerical tasks across desktop, web, and mobile platforms. This tool will facilitate tasks such as drafting documents and spreadsheets.
The launch of GPT-5.6 comes amid competitive releases from rivals including SpaceXAI and Meta. Emphasizing its performance, OpenAI positioned GPT-5.6 as a direct challenge to its main competitor, Anthropic.
Using the Artificial Analysis Coding Agent Index, OpenAI stated that Sol is the best coding model to date, surpassing Anthropic’s Fable 5 by 2.8 points with a score of 80. Sol reportedly uses fewer than half the output tokens, takes less than half the time to execute tasks, and costs approximately one-third less than Fable 5.
According to OpenAI, Terra outperforms Fable 5, while Luna exceeds the performance of Opus 4.8. GPT-5.6 is now available via ChatGPT, Codex, and the OpenAI API. Pricing for token usage is set at $5 input and $30 output for Sol, $2.50 input and $15 output for Terra, and $1 input and $6 output for Luna.
