Sol and Luna pricing and benchmarks compared with Opus 5.5, two hours after launch
A post says Sol is priced at $2/$10, half of Opus 5.5, and Luna at $0.10/$0.50, while their AA Intelligence Index scores stay roughly level with the previous generation.
Anthropic drops Opus 5.5. Barely two hours later, OpenAI drops GPT-6 Sol and Luna. We’re having one of the biggest weeks in AI, and Ai labs are not slowing down. For context; Sol is built for complex coding and agentic workflows, while Luna is the cheaper model for everyday, high volume tasks. And OpenAI really went crazy with the pricing: ⟢ GPT6 Sol costs $2/$10 in/out tokens. That’s half the price of Opus 5.5. ⟢ Luna is even cheaper at $0.10/$0.50. OpenAI also claims Sol makes around half as many factual mistakes as GPT-5.6 Sol. Early AA benchmarks are also pretty interesting. ➥ Sol scored 47 on the Intelligence Index, roughly the same as GPT-5.6 Sol. ➥ Luna also scored 37, around the same as its predecessor. So we’re getting significantly cheaper models without a massive jump in their overall intelligence scores. And tbh, I’m pretty happy with that. I’m curious to see how Sol holds up against Opus 5.5 in actual sessions. Will be trying it out on Devin & codex, and I’ll let you guys know what I think.
This post quotes another post: “OpenAI releases GPT-6 Sol and Luna with 50% lower API prices”