OpenAI发布GPT-6.1 Sol提升成本效率
OpenAI新发布的GPT-6.1 Sol成本大幅降低，同时修复了GPT-6 Luna的视觉理解问题，让模型在多个基准测试中表现提升。
OpenAI推出GPT-6.1 Sol模型，在最大性能下每项智能指数任务成本仅0.72美元，比GPT-6 Astra低75%。该模型在所有性能水平上都突破了成本效率帕累托前沿。同时，OpenAI修复了GPT-6 Luna的多模态输入问题，使其在GDP.pdf等视觉理解基准上得分提升。
GPT-6.1 Sol pushes the cost efficiency frontier for OpenAI models. OpenAI has also fixed an issue reading multimodal input which improves Intelligence Index results for GPT-6 Luna by 1 point.
At max effort, GPT-6.1 Sol costs less than a quarter of GPT-6 Astra per Intelligence Index task ($0.72 vs $3.26). It also costs 31% less per task than GPT-6 Sol ($1.04) and 64% less than GPT-5.6 Sol ($1.99). All effort levels of GPT-6.1 Sol push out the cost efficiency Pareto frontier: for a given level of intelligence, there is no cheaper model.
OpenAI also fixed a bug in image encoding for GPT-6 Luna and Sol. GPT-6 Luna (max) has gained a point in the Intelligence Index, with improvement in evals which benefit from visual understanding such as GDP.pdf (2.4pt gain), MMMU-Pro (4.1pt gain), GDPval-AA v2.1 (71 Elo gain), AA-Briefcase (36 Elo gain). GPT-6 Sol shows negligible improvement.
