WHAT NEW IN X10.31 SAT

X 上的 AI 合集档案 · 22 个合集 · 61,986 条

gpt-6IMAGE

GPT-6 Sol 与 Luna 规格:105 万上下文、定价与用途

配图帖列出 Sol/Luna 分工、1.05M 上下文、128K 输出与 API 价格,并说明缓存读取打九折。

Abhishek Yadav
@yabhishekhd2026.09.24收录原推发布于 2026.09.23
GPT-6 Sol 与 Luna 规格:105 万上下文、定价与用途
Abhishek Yadav @yabhishekhd原推发布于 2026.09.23
在 X 上打开原推

After Astra, now two GPT-6 models arrived: Sol and Luna ⚡ Sol is for doing coding work and agents. Luna is for high-volume tasks like extraction and routing. Extraction is like it reads a huge number of documents and gives the information we exactly need in a hurry, removing the clutter or unuseful info. Routing is like assigning a specific task to the guy who performs that best. Major changes: 1. Both get a 1.05M-token context window 2. Up to 128K output tokens 3. Sol: $2/M input, $10/M output 4. Luna: $0.10/M input, $0.50/M output 5. Cached reads get a 90% discount (sending the same document or same text again now costs less) 6. Available in the API, ChatGPT Work and Codex. Free and Go users can use Luna in the desktop app #OpenAI #GPT6 #Developers

x.com/yabhishekhd/status/2102554659894579690采集于 2026.09.24 10:07