引用OpenAI
OpenAI开始有限预览GPT‑5.6系列,包括旗舰模型Sol、日常使用的均衡模型Terra以及快速经济的Luna。Terra性能与GPT‑5.5相当,但成本降低一半。定价按每百万token计算:Sol输入5美元/输出30美元,Terra输入2.50美元/输出15美元,Luna输入1美元/输出6美元。新版本还引入了更可预测的提示缓存,支持显式缓存断点和最低30分钟缓存生命周期,缓存写入按输入价格的1.25倍计费,缓存读取享受90%折扣。
OpenAI开始有限预览GPT‑5.6系列,包括旗舰模型Sol、日常使用的均衡模型Terra以及快速经济的Luna。Terra性能与GPT‑5.5相当,但成本降低一半。定价按每百万token计算:Sol输入5美元/输出30美元,Terra输入2.50美元/输出15美元,Luna输入1美元/输出6美元。新版本还引入了更可预测的提示缓存,支持显式缓存断点和最低30分钟缓存生命周期,缓存写入按输入价格的1.25倍计费,缓存读取享受90%折扣。
METR conducted a predeployment evaluation of GPT-5.6 Sol, assessing its capabilities and risks before release. The evaluation focused on the model's performance in areas such as autonomous task completion and potential misuse. Findings informed safety decisions ahead of deployment.
OpenAI announced a limited preview of its GPT‑5.6 series but is releasing them gradually because the U.S. government requested a staggered rollout, with customer-by-customer approval. Commerce Secretary Howard Lutnick later cautioned against launching without broader agency sign-offs.
OpenAI's GPT-5.6 Sol was evaluated against standard cybersecurity benchmarks, showing improved performance in vulnerability detection and threat analysis compared to previous models, though it still struggles with highly specialized attack vectors and adversarial inputs.