OpenAI launches GPT-6 Sol and Luna, cutting API prices in half
Built on the GPT-6 Astra work, Sol targets everyday demanding tasks and Luna targets high-volume ones -- OpenAI says Sol makes about half as many factual mistakes as its predecessor at a fraction of the cost.
OpenAI released two new models on September 22: GPT-6 Sol and GPT-6 Luna, both built on the technology behind GPT-6 Astra (OpenAI's frontier release earlier in September) but repackaged as faster, cheaper alternatives rather than a new flagship tier. API pricing is roughly half of GPT-5.6's promotional rate: $2/$10 per million input/output tokens for Sol, and $0.10/$0.50 for Luna.
Two models, two jobs
Sol is aimed at demanding everyday work -- OpenAI's own framing is that it reaches "Astra-level reliability at much lower cost." Luna is aimed at high-volume tasks where per-token cost dominates the decision. OpenAI's factuality claim -- that Sol makes about half as many mistakes as its predecessor -- comes from the company's internal evaluation built on de-identified real-world conversations where users flagged errors, which is a real methodology but not an independent benchmark; treat the specific "half as many" figure as OpenAI's own claim rather than externally verified.
Availability
Both models are live now for Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, with API access enabled at launch; free and Go-tier users get GPT-6 Luna in the desktop app. Combined with Anthropic's Opus 5.5 price cut the same day, the pattern this week is the same on both sides: frontier labs pushing near-flagship quality down into their cheaper tiers, rather than only pushing the top tier's capability up.