🤖 OpenAI Releases GPT-6 Sol and GPT-6 Luna — Twice as Cheap as Predecessors
OpenAI has expanded the GPT-6 lineup with two models: GPT-6 Sol and GPT-6 Luna, built on the foundation of the flagship GPT-6 Astra, which was released earlier in September 2026. GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens — exactly 50% cheaper than GPT-5.6 Sol ($4/$20). GPT-6 Luna is $0.10/$0.50 per 1M tokens, compared to $0.20/$1.20 for GPT-5.6 Luna: input is 50% cheaper, output is approximately 58% cheaper. OpenAI confirmed that the prices are fixed, not promotional. The savings were achieved through inference optimization and caching: the discount on reading cached input tokens has increased to 90%, and a Prompt Caching Dashboard and a tool for diagnosing missed cache have been introduced.
🌍 OpenAI has shifted price competition to the mid-tier and mass-market segments. GPT-6 Sol launched at exactly the price of Claude Sonnet 5 ($2/$10) and twice as cheap as Claude Opus 5.5 ($4/$20), which Anthropic released the same morning. Luna's input price ($0.10 per 1M) is nearly on par with open MIT-licensed models like Xiaomi MiMo-V2.6-Flash ($0.14/$0.28). A permanent, not promotional, 50% price cut plus a 90% discount on cached tokens changes the economics of agentic and automation workloads in production.
👤 The models are already available via API: Sol is suitable for code review, debugging, and data analysis, while Luna is for mass summarization and information extraction. The stated figures should be verified on your own tasks: GPT-6 Sol (xhigh) scores 33.2% on AutomationBench 1.0.6, compared to 26.9% for Claude Opus 5 (max), at a task cost of about $0.27 — approximately 9% of the competitor's price, 56.4% on Agents' Last Exam V1, and 68.8% on DeepSWE v1.1 (max), trailing only 1.1 percentage points behind the best result of Claude Fable 5 (69.9%) at a task price about 80% lower. Luna at high effort levels is close to GPT-5.6 Sol at approximately 1/100 of its price. Those who count tokens should recalculate their automation budget taking the new caching into account: GitHub reported a reduction of more than 50% in the share of non-cached prompt tokens across billions of requests.
Source 1: https://openai.com/index/introducing-gpt-6-sol-and-luna/