GPT-6 Luna

GPT-6 Luna shipped alongside GPT-6 Sol on September 22, 2026 as gpt-6-luna, and OpenAI describes it as its most efficient model for focused, high-volume tasks. Pricing is $0.10 per million input tokens, $0.01 cached, and $0.50 output, down from $0.20 and $1.20 on GPT-5.6 Luna, so the output cut is closer to 58 percent than the 50 percent OpenAI's launch table rounds it to. Prompts above 272,000 input tokens bill the whole request at $0.20, $0.02, and $0.75. Context is 1,050,000 tokens with 128K output, text and image input, and the same reasoning effort ladder as Sol from none through max. OpenAI's claim is that Luna matches GPT-5.6 Sol at about a hundredth of its cost, which is a vendor comparison worth testing on your own tasks. The independent numbers are more measured and still strong for the price: Cognition puts it at 42.4 percent on FrontierCode v1.1 at max effort, the Vals Index at 58.45 percent, twentieth of 65, and Artificial Analysis at 37. On ChatGPT, Free and Go users get Luna in the desktop app, and Codex now suggests switching to it when a user nears a rate limit. OpenAI did not publish SWE-bench Verified, MMLU-Pro, GPQA Diamond, BrowseComp, or HLE figures for it.