GPT-5.6 Luna is the fastest and most cost-efficient model in OpenAI's GPT-5.6 family, sitting below Terra and the flagship Sol model in capability. It is designed for high-volume, latency-sensitive workloads such as chat, classification, extraction, tagging, batch drafting, and well-scoped coding tasks, where throughput and cost matter more than deep, long-horizon reasoning.
Architecture and Capabilities
Luna shares the same 1.05 million token context window and 128,000 token maximum output as Sol and Terra, and accepts both text and image inputs. It supports tool use, function calling, structured outputs, and prompt caching, with configurable reasoning effort levels. Luna is accessed via the API using the gpt-5.6-luna model identifier; the general gpt-5.6 alias routes to Sol rather than Luna.
Performance
OpenAI describes Luna as nearly matching the peak performance of the prior-generation GPT-5.5 at well under half the estimated API cost. On agentic coding evaluations such as Terminal-Bench 2.1, independent roundups have reported Luna scoring competitively for its tier, in some cases ahead of the mid-tier Terra model, while trailing Sol. Document-processing benchmarks have similarly found that Luna gives up little accuracy relative to Sol on text- and table-heavy content despite its substantially lower cost, though the broader GPT-5.6 family shows reduced accuracy on chart- and layout-heavy documents.
Safety
Luna was released alongside Sol and Terra as part of OpenAI's coordinated GPT-5.6 rollout, which included a limited preview period tied to engagement with the U.S. government over the family's cybersecurity and biological-risk capabilities. Under OpenAI's Preparedness Framework, GPT-5.6 models, including Luna, are treated as High capability in cybersecurity and biological/chemical risk domains, without reaching the framework's Critical threshold.
Exact parameter counts and training architecture details have not been publicly disclosed.