OpenAI expanded its GPT-6 lineup on Tuesday by releasing the lower-cost Sol and Luna models, slashing API pricing by 50% or more compared to previous promotional rates. The new models bring reasoning, coding, and computer-use capabilities closer to the flagship Astra tier while targeting everyday enterprise tasks at lower operating costs.
Pricing Structures and API Cost Reductions
The core strategy behind the expansion centers on inference efficiency and caching improvements. According to the company, these engineering upgrades allow OpenAI to pass direct savings to developers by reducing API rates to half of what their GPT-5.6 promotional pricing previously cost. An OpenAI spokesperson confirmed to VentureBeat that these newly announced rates represent permanent pricing rather than temporary introductory discounts.
Under the updated fee schedule, GPT-6 Sol is priced at $2 per million input tokens and $10 per million output tokens. Meanwhile, GPT-6 Luna drops to $0.10 per million input tokens and $0.50 per million output tokens. Against earlier generation rates, the drop is substantial; GPT-5.6 Sol currently costs $4 per million input tokens and $20 per million output tokens, making the new Sol model exactly 50% cheaper in both directions.
Intended Roles for Sol and Luna in Enterprise Workflows
While the flagship GPT-6 Astra model remains reserved for the most complex, multi-faceted projects and heavy research tasks, Sol and Luna fill distinct operational niches for daily software development and high-volume clerical duties. Sol is engineered for iterative coding tasks, debugging, data analysis, and complex work workflows that demand careful, multistep validation. Luna functions as the lightweight, high-volume option designed for repetitive, fixed-goal jobs such as document summarization, extraction, and answering straightforward questions.
Inside OpenAI, internal usage has surged as coding agents take on longer and more demanding tasks. Valued at API prices, daily token usage has exceeded $600 for the median researcher and $7,000 for researchers at the 90th percentile. Deploying lower-cost models gives engineering teams the financial headroom to let automated coding tools operate continuously without straining budgets.
Ecosystem Integration and GitHub Copilot Rollout
Developer adoption is already expanding beyond OpenAI’s native API channels.
- GPT-6 Sol is accessible to Copilot Pro+, Max, Business, and Enterprise plans.
- GPT-6 Luna is available across Copilot Pro, Pro+, Max, Business, and Enterprise plans.
- Supported environments include Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot cloud agent, GitHub Mobile for iOS and Android, JetBrains, Xcode, and Eclipse.
Enterprise and business plan administrators can manage access through their default model policy settings, where new models are enabled automatically unless explicitly toggled off.
Competitive Pressures in the Frontier Model Market
The timing of the release underscores intense rivalry across the artificial intelligence sector. Sol lands directly at Claude Sonnet 5 pricing—matching Anthropic’s permanent $2/$10 rate—while undercutting newer competitors. Anthropic shifted the landscape on the same morning by introducing Claude Opus 5.5 at $4 input and $20 output per million tokens, leaving GPT-6 Sol 50% cheaper than Opus 5.5 on raw token pricing.
Meanwhile, Google maintains an aggressive posture at the flash tier with Gemini 3.8 Flash priced at $0.75 input and $3.75 output under promotional terms running through the end of the year. As enterprises look closer at the unit economics of multi-step agent workflows, the battle among foundation model providers has shifted decisively from raw capability demonstrations toward cost efficiency at scale.
