OpenAI has officially broadened its latest model generation by introducing GPT-6 Sol and GPT-6 Luna. Following the initial debut of its top-tier GPT-6 Astra flagship—which has already seen rapid adoption across workflow tools like Hex's automated data platforms and Higgsfield's AI video generation pipelines—the artificial intelligence firm is distributing its current intelligence capabilities across more budget-conscious price tiers. The two newly added releases aim to balance operational expense with high-level performance across enterprise, programming, and agentic workflows.
Expanding the GPT-6 Ecosystem
While GPT-6 Astra remains OpenAI's primary option for complex reasoning and deep constraint analysis, Sol and Luna target distinct balance points along the capability-to-cost spectrum. Both models were trained using alignment techniques and architectural methods similar to Astra. As a result, they incorporate family-wide upgrades in factual reliability, long-context understanding, and multi-step execution while offering a clearer, more concise output style in technical conversations.
Both Sol and Luna sport a context length of 1,050,000 tokens and support max outputs of 128,000 tokens, fully integrating with OpenAI's core agent capabilities like computer use, web search, and function calling. For teams navigating frequent model announcements, our guide on how to read a model release without being sold to highlights why context window size and token pricing are often far more critical for production systems than benchmark scores alone.
Pricing Cuts and Tier Capabilities
The headline development surrounding this rollout is the dramatic drop in deployment economics. Thanks to backend advances in system inference and prompt caching, OpenAI has halved the API pricing for both Sol and Luna compared to previous GPT-5.6 promotional levels.
- GPT-6 Sol: Positioned directly below Astra, Sol serves as the mid-tier option for complex coding, document evaluation, and judgment-heavy business automation. Standard API access is priced at $2.00 per million input tokens and $10.00 per million output tokens.
- GPT-6 Luna: Built as a lightweight, highly responsive option, Luna handles high-volume tasks such as text categorization, structured data extraction, and routine chat processing. Standard API rates are set at $0.10 per million input tokens and $0.50 per million output tokens.
On AutomationBench, GPT-6 Sol at max effort setting outperforms Anthropic's Claude Opus 5 while costing only 9 percent as much per task. These pricing adjustments reflect broader industry shifts toward usage-based models, much like the recent transition seen when GitHub Copilot shifted to usage-based AI credits.
Distribution and Deployment
GPT-6 Sol and Luna have rolled out into the OpenAI API alongside availability in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu subscription tiers. Furthermore, free users gain access to Luna through OpenAI's desktop application, making the updated architecture immediately accessible across consumer and enterprise environments alike. As developers integrate these models into daily operations, these lightweight options are expected to rank alongside the best AI productivity and workflow tools for professionals.
