爆料：GPT-6 Sol 与 Luna 定位与定价细节
帖子称 Sol 面向复杂编程与专业工作、Luna 主打超低价日常任务，并列出两者每百万 token 的输入输出价格。
GPT-6 Sol and Luna are MUCH more interesting than just “cheaper GPT-6 models.” OpenAI is bringing a huge portion of Astra’s advances in reasoning, factual reliability, coding, computer use, and alignment down into models designed for everyday work and massive-scale deployment. The positioning is now very clear: → GPT-6 Astra: maximum capability, hardest tasks → GPT-6 Sol: complex coding + professional work at dramatically lower cost → GPT-6 Luna: fast, absurdly cheap everyday work at scale And the pricing is genuinely insane: GPT-6 Sol → $2.00 / 1M input → $10.00 / 1M output GPT-6 Luna → $0.10 / 1M input → $0.50 / 1M output That’s 50% below the promotional pricing of their GPT-5.6 counterparts. The official evaluations also show the same basic story: Astra remains the ceiling, but Sol brings much of that generation’s capability into a completely different price/performance regime, while Luna is targeting high-volume workloads at almost ridiculous prices. This may be the most important part of GPT-6: not just making frontier intelligence better, but making it CHEAP.
这条是引用着另一条写的：「OpenAI 提高用量上限并降价」