OpenAI slashes API costs with GPT-6 Sol and Luna
OpenAI has halved API costs for its new GPT-6 Sol and Luna models, escalating an industry price war to win over enterprise developers facing cheaper open-source alternatives.

On September 22, 2026, OpenAI launched GPT-6 Sol and GPT-6 Luna, cutting prices by 50 percent compared to their predecessors, GPT-5.6 Sol and GPT-5.6 Luna. The new GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens. Meanwhile, the smaller GPT-6 Luna is priced at $0.10 per million input tokens and $0.50 per million output tokens. Both models were trained using methods similar to those used for GPT-6 Astra, which debuted earlier in the month.
This aggressive pricing strategy directly targets rivals like Anthropic and Google. OpenAI claims that GPT-6 Sol matches Anthropic’s Claude Fable 5.1 in coding capabilities but at a fraction of the cost; Claude Fable 5.1 currently charges $10 per million input tokens and $50 per million output tokens. Google is also in the mix, offering promotional rates on Gemini 3.8 Flash through December at $0.75 for input and $3.75 for output per million tokens.
For enterprise developers and AI practitioners, these dramatic price cuts make deploying frontier-class models significantly more viable for high-volume, daily workflows. Industry analysts note that the price cuts are a direct response to pressure from highly capable, cheaper open-source models, particularly those originating from China. While some experts question whether such steep discounts are sustainable over the long term, the immediate result is a massive financial relief for teams building agentic workflows and complex applications.
Beyond immediate adoption, the pricing maneuver supports OpenAI's broader business goals as the company moves toward a potential public listing, having made its initial IPO filing in June. By lowering the barrier to entry, OpenAI aims to rapidly scale usage of its GPT-6 ecosystem. As Futurum Group analyst Bradley Shimmin noted, this intense competition forces closed-model developers to prioritize practical enterprise utility and security over mere valuation metrics.
This is our own summary of reporting by AI Business


