OpenAI Previews GPT-5.6 Series: Sol, Terra, and Luna Models with Pricing and Caching Updates
English summary
OpenAI is previewing the GPT-5.6 series, comprising the flagship Sol, balanced Terra, and low-cost Luna models. Terra matches GPT-5.5 performance at half the cost, while Luna provides strong capability at the lowest price. The models are available in limited preview to trusted partners, with general availability planned in coming weeks. Token pricing per 1M tokens is $5/$30 for Sol, $2.50/$15 for Terra, and $1/$6 for Luna. The release introduces predictable prompt caching with explicit breakpoints and a 30-minute minimum cache life; cache writes cost 1.25x the model’s input rate, and cache reads retain a 90% discount.
Chinese summary
OpenAI 预览了 GPT-5.6 系列，包括旗舰模型 Sol、均衡型 Terra 和低成本模型 Luna。Terra 以一半的成本实现了与 GPT-5.5 相当的性能，Luna 则提供了最低成本下的强大能力。这些模型目前向受信任的合作伙伴开放有限预览，计划在未来几周内全面推出。每百万代币的定价为：Sol 5/30 美元，Terra 2.50/15 美元，Luna 1/6 美元。该版本还引入了带有显式断点和 30 分钟最低缓存生命周期的可预测提示缓存；缓存写入按模型输入费率的 1.25 倍计费，缓存读取则保留 90% 的折扣。
Key points
GPT-5.6 series includes three models: Sol (flagship), Terra (balanced, 2x cheaper than GPT-5.5), and Luna (fastest, cheapest).
GPT-5.6 系列包含三款模型：旗舰 Sol、均衡型 Terra（比 GPT-5.5 便宜一倍）以及最快最便宜的 Luna。
The models are in limited preview with trusted partners, with general availability expected in coming weeks.
模型正向受信任合作伙伴提供有限预览，计划在未来几周内全面推出。
Per 1M token pricing: Sol $5 input / $30 output, Terra $2.50 / $15, Luna $1 / $6.
每百万代币定价：Sol 输入 $5 / 输出 $30，Terra $2.50 / $15，Luna $1 / $6。
Terra offers GPT-5.5 competitive performance at 2x lower cost.
Terra 以降低一半的成本提供与 GPT-5.5 相当的性能。
New predictable prompt caching: explicit breakpoints, 30-minute minimum cache life, cache writes billed at 1.25x input rate, cache reads get 90% discount.
新的可预测提示缓存：显式断点、30 分钟最低缓存生命周期，缓存写入按输入费率的 1.25 倍计费，缓存读取享受 90% 折扣。