OpenAI releases the GPT-5.6 model family in Luna, Terra and Sol variants

OpenAI/GPT models

OpenAI made GPT-5.6 generally available in three tiers: Sol (flagship), Terra (everyday work) and Luna (fast, cheap). The number marks the generation while capability tiers can advance independently. Sol scored 53.6 on OpenAI's Agents' Last Exam benchmark, 13.1 points above Anthropic's Claude Fable 5.

• Three-tier model family: Luna ($1/$6 per 1M input/output tokens), Terra ($2.50/$15) and Sol ($5/$30); all with 1.05M-token context.
• Agents' Last Exam evaluates professional workflows across 55 fields; Sol scores 53.6 vs. Claude Fable 5 at 40.5.
• Ships programmatic tool calling, subagents and explicit prompt-cache control for long-running agent workloads.