OpenAI expands GPT-6 lineup with Sol and Luna at half the price

OpenAI has launched GPT-6 Sol and GPT-6 Luna, expanding its latest model family with faster and more affordable options following the release of GPT-6 Astra earlier this month.

OpenAI said API prices for both models are 50% lower than the promotional pricing of their GPT-5.6 predecessors, helped by improvements in inference and caching.

GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens, down from $4 and $20 for GPT-5.6 Sol. GPT-6 Luna costs $0.10 per million input tokens and $0.50 per million output tokens.

The models use similar training methods to GPT-6 Astra, with improvements spanning professional work, factuality, coding, computer use and alignment. Astra remains OpenAI’s most capable model for workloads where maximum performance matters.

On AutomationBench, GPT-6 Sol at xhigh effort scored 33.2%, ahead of Claude Opus 5 at max effort and Claude Fable 5.1 with Opus fallback, while OpenAI said it completed tasks at substantially lower cost.

Coding performance also improved. GPT-6 Sol matched Claude Fable 5.1 on FrontierCode at a lower cost and scored 68.8% on DeepSWE 1.1, within 1.1 percentage points of Fable’s highest reported score while costing about 80% less per task, according to OpenAI.

GPT-6 Luna is aimed at even lower cost workloads. On DeepSWE, Luna scored 66.6% at maximum effort, with OpenAI estimating task costs 93% below Claude Opus 5 and 96% below Claude Fable 5 at the compared settings.

OpenAI also said GPT-6 Sol produces about half as many factual errors as GPT-5.6 Sol on an internal evaluation based on difficult conversations where users previously flagged mistakes.

The company has also improved prompt caching across GPT-6, offering discounts of 90% on cached input token reads and higher cache hit rates for long conversations and agents.

GPT-6 Sol and GPT-6 Luna are available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users. GPT-6 Luna is also available to Free and Go users through the desktop app, while both models are available through the API.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.