OpenAI has made its latest flagship model series, GPT-5.6, generally available in three sizes: Luna, Terra, and Sol. The release introduces new API capabilities including programmatic tool calling, multi-agent support, and explicit prompt cache breakpoints.
- Pricing is set at $1/$6 for Luna, $2.50/$15 for Terra, and $5/$30 per 1M input/output tokens respectively.
- On the Agents’ Last Exam benchmark, GPT-5.6 Sol achieved a score of 53.6, outperforming Claude Fable 5 by 13.1 points.
- OpenAI claims GPT-5.6 Terra and Luna outperform Fable 5 at roughly one-sixteenth the cost on long-running professional workflows.
- New API features allow composing JavaScript for tool calls, spinning up subagents for parallel work, and setting explicit prompt cache breakpoints.
The models aim to provide more useful work per token, with OpenAI highlighting efficiency gains across different reasoning efforts.