GPT-6 Sol vs Luna: Amazon Bedrock Pricing & Key Differences

Overview of GPT-6 Sol and Luna Availability

OpenAI has released GPT-6 Sol and GPT-6 Luna, expanding the GPT-6 family of models. These two models are now generally available on Amazon Bedrock, giving users more options to match intelligence and efficiency to each workload. The GPT-6 family now includes GPT-6 Astra, GPT-6 Sol, and GPT-6 Luna. The launch of GPT-6 Sol and Luna follows the rollout of GPT-6 Astra, which was released earlier this month. This sequence of releases allows developers and organizations to select from a broader range of models, each designed for different performance and cost requirements.

Pricing and Cost Structure

OpenAI has set the pricing for GPT-6 Sol and GPT-6 Luna at levels significantly lower than their predecessors. GPT-6 Sol is priced at $2 per 1 million input tokens and $10 per 1 million output tokens. GPT-6 Luna is priced at $0.10 per 1 million input tokens and $0.50 per 1 million output tokens. OpenAI states that GPT-6 Sol is exactly 50% cheaper than GPT-5.6 Sol in both input and output token costs. OpenAI also states that GPT-6 Luna is 50% cheaper on input and 58.3% cheaper on output compared to GPT-5.6 Luna. These reductions are not promotional or introductory rates; OpenAI has confirmed that these are permanent prices. The company is reducing the API price for GPT-6 Sol and Luna by 50%, making both models more accessible for a wider range of use cases. This pricing structure is intended to provide sustainable, lower-cost options for users who require advanced AI capabilities but need to manage operational expenses. OpenAI notes that the GPT-6 models lead across the cost and intelligence curve, combining capabilities at every tier with infrastructure that delivers them efficiently at scale.

Target Workloads, Model Design, and Technical Improvements

GPT-6 Sol and GPT-6 Luna are mid-priced and lower-tier models designed for enterprise tasks. OpenAI describes these models as enterprise workhorses for common tasks that require less intelligence than the top-tier Astra model. GPT-6 Sol is designed for complex coding and agentic workflows, making it suitable for developers and knowledge workers who perform tasks such as building features, reviewing code, debugging, and analyzing data. GPT-6 Luna is designed for focused, high-volume tasks, providing a cost-effective solution for workloads that prioritize efficiency and throughput. Both models were designed to offer lower-cost options for professional work, coding, automation, and computer-use tasks.

OpenAI trained Sol and Luna using methods similar to those used for Astra, including improvements in reasoning, factual reliability, coding, computer use, and alignment. According to OpenAI, improvements in caching and inference allow the company to serve these models at lower cost, and the savings are being passed directly to users and customers. These technical improvements contribute to the models’ ability to deliver strong performance for a variety of professional and enterprise tasks, while maintaining cost efficiency.

Ecosystem Rollout and Availability

The rollout of GPT-6 Sol and Luna extends to multiple platforms. Both models are now generally available on Amazon Bedrock. OpenAI is also rolling out GPT-6 Sol and Luna to ChatGPT Work and Codex for Plus, Pro, Business, and Enterprise customers. In addition, both models are available to Edu users. Free and Go users can access GPT-6 Luna through the desktop app. However, the models are not yet available in the Chat interface itself. OpenAI has stated that the rollout to Work and Codex will proceed gradually to preserve service stability. Availability may depend on rollout timing and workspace settings, and Enterprise administrators must enable the new models for their organizations. This staged rollout approach is intended to ensure a smooth transition and reliable service for all users.

Capability Comparison and Positioning

GPT-6 Sol and Luna are more capable than the GPT-5.6 versions they replace. OpenAI describes these two newest models as offering improved performance across a range of tasks, while remaining more affordable than Astra and the GPT-5.6 versions. GPT-6 Astra remains OpenAI’s top-tier option for workloads where maximizing capability is the primary concern. By contrast, GPT-6 Sol and Luna are positioned to provide efficient and cost-effective solutions for users who need to allocate different classes of work across a portfolio of models. The introduction of these models reflects OpenAI’s approach to offering a range of options that allow organizations to match model selection to their specific needs and budgets.

Summary

The launch of GPT-6 Sol and Luna marks an expansion of OpenAI’s GPT-6 lineup, following the earlier release of GPT-6 Astra. With their lower pricing, OpenAI states that GPT-6 Sol is exactly 50% cheaper than GPT-5.6 Sol, and GPT-6 Luna is 50% cheaper on input and 58.3% cheaper on output compared to GPT-5.6 Luna. These models are now available on Amazon Bedrock, ChatGPT Work, Codex, and the desktop app for eligible users. Designed as mid-priced and lower-tier options, GPT-6 Sol and Luna are intended for a range of enterprise and professional workloads, offering improved capabilities and cost efficiency. OpenAI’s approach allows organizations and developers to select the model that best fits their requirements, balancing performance and budget across the GPT-6 family.