OpenAI launches GPT-6 Sol and Luna at ~50% cheaper than 5.6, with stronger coding
charliermarsh · x · 2026-09-23
OpenAI introduced GPT-6 Sol and Luna, bringing Astra advances to faster, cheaper models priced roughly 50% below GPT-5.6. Sol at xhigh effort beats Claude Opus 5 at max effort on AutomationBench at 9% of the per-task cost; Luna at max effort exceeds GPT-5.6 Sol at medium effort on OSWorld 2.0 offline at one-tenth the cost. Sol makes about half as many factuality errors as 5.6 Sol on internal evals, and improved prompt caching gives 90% discounts on cached input-token reads for agents.
Related event: OpenAI Launches GPT-6 Sol and Luna with 50% API Price Cut(6 posts)→
More from Models
- Matt Shumer launches first Opus 5.5 Gauntlet Loop, live-testing the model to redesign his site — mattshumer_ · 2026-09-23
- Codex users say weekly limits burned in 2 days, questioning OpenAI's '50% cheaper' Sol/Luna pricing — cocktailpeanut · 2026-09-23
- Two studies: AI chatbots get financial advice wrong 57%-85% of the time — PtrPomorski · 2026-09-23
- Early GPT-6 Sol access: 1-hour tasks done in 20 minutes with half the tokens — flavioAd · 2026-09-23
- GPT-6 Sol lands in Perplexity as Computer's default Light option — perplexity_ai · 2026-09-23
- Opus 5.5 medium costs slightly more but beats Astra on nearly everything — ChrisGPT · 2026-09-23