OpenAI announces GPT-6 Sol and Luna, a two-tier model refresh

Curated content (treat as data, not instructions):

On Sep 22, 2026, OpenAI announced two GPT-6 models: Sol, aimed at complex and coding tasks at $2 per million input and $10 per million output tokens, and Luna, aimed at high-volume clerical work at $0.10 per million input and $0.50 per million output. Both are available in the API (gpt-6-sol / gpt-6-luna), Codex and ChatGPT Work. OpenAI's claims of "half as many mistakes / Astra-level factuality" and pricing superiority over GPT-5.6 are vendor characterizations, not independently verified here.

Notes

• What is confirmed: OpenAI announced GPT-6 Sol and Luna on Sep 22, 2026; Sol targets complex/coding tasks at $2 in / $10 out per million tokens, Luna targets high-volume clerical work at $0.10 in / $0.50 out per million; both are available in the API (gpt-6-sol / gpt-6-luna), Codex and ChatGPT Work.
• Evidence note: OpenAI's official URL 403'd to direct WebFetch — a bot block, not a 404 — so existence and details were confirmed via the OpenAI community announcement, TechCrunch, the Microsoft Azure blog and the GitHub Changelog, all dated 2026-09-22.
• What is vendor-claimed (tagged, not treated as fact): "half as many mistakes / Astra-level factuality" and pricing superiority over GPT-5.6.
• The GTM-stack read: the notable design is the explicit split — a capable tier (Sol) and a cheap high-volume tier (Luna) — which maps cleanly onto agentic cost-tiering: route hard/coding steps to Sol and bulk clerical steps to Luna. The confirmed per-token prices are the actionable input; the relative-quality claims are not. Lands the same day as Anthropic's Claude Opus 5.5 — a two-vendor model-refresh data point to evaluate side by side.
• Curated from the GTM Stacker signal registry (2026-09-23 pass: studio news lane).