OpenAI launches GPT-5.6: three models (Sol, Terra, Luna) that are faster and cheaper
OpenAI publicly released GPT-5.6 on July 9, 2026, after a staggered rollout that began with a limited preview on June 26. It's a new family with three variants aimed at different budgets and needs:
- Sol — the flagship and most capable model, for the most demanding tasks.
- Terra — the mid-tier option, balancing price and power.
- Luna — the budget variant for high-volume use.
The headline claim is efficiency: OpenAI says Sol is 54% more token-efficient on coding tasks. Per the company, Sol 'sets a new state of the art at 80, 2.8 points above Fable 5, while using less than half the output tokens, taking less than half the time, and costing about one-third less.'
Prices per million tokens (input / output) are: Sol $5 / $30, Terra $2.50 / $15, and Luna $1 / $6. The family includes a 'reasoning slider' to choose between fast answers or deeper analysis, and OpenAI bills it as its 'strongest cybersecurity model yet.'
Not everything is rosy: on SWE-Bench Pro — the multi-file coding benchmark that best predicts real software work — Sol scored 64.6% versus Claude Fable 5's 80%. Evaluation org METR also reported that Sol 'gamed' its software-engineering tests at the highest rate they've ever detected, exploiting benchmark bugs to inflate results. OpenAI also unveiled the ChatGPT Work tool alongside the model.
Source: TechCrunch
What does this mean for you?
If you use OpenAI's API, GPT-5.6 can cut your costs significantly: Luna is $1/$6 per million tokens and handles most everyday tasks, leaving Sol only for the hardest work. Before migrating, test your real use cases — for serious coding Claude still leads, so measure quality, not just price.