OpenAI announced GPT-6 Sol and Luna on September 22, 2026, giving users more options to choose an AI model that fits their needs and budget. The models can handle tasks such as coding, research, writing, and computer-based work while making AI more affordable. On October 7, 2026, OpenAI expanded GPT-6 to ChatGPT users, replacing GPT-5.6 Sol and Luna in the Chat experience.
Table of Contents
What Are GPT-6 Sol and Luna?
After launching GPT-6 Astra, OpenAI introduced GPT-6 Sol and Luna as cost-effective options for users who need capable AI models. Both models aim to balance AI performance with the cost of running tasks. Sol is designed for complex work, including demanding coding projects, research, and professional workflows. Luna focuses on affordability and efficiency, making it suitable for everyday questions and applications that process large numbers of requests.
Both models can write content, summarize information, analyze documents, and help solve problems. The difference isn’t that Luna can only do simple tasks or that Sol is exclusively for professionals. The main difference is how they balance performance and cost.
OpenAI has also introduced Intelligent UI in ChatGPT, which can turn answers into interactive charts, diagrams, tables, and calculators. These elements make complex information easier to understand. GPT-6.1 Sol was also introduced after the original GPT-6 Sol and Luna launch. This newer model aims to deliver capabilities closer to GPT-6 Astra while costing less.
GPT-6 Sol vs GPT-6 Luna: Key Differences
GPT-6 Sol and Luna mainly differ in cost and capabilities. The key differences are listed below.
| Model | GPT-6 Sol | GPT-6 Luna |
| Use | Complex and professional work | Everyday tasks |
| Coding | Complex coding tasks | Everyday coding and assistance |
| Research | Complex research and analysis | Summarization and everyday research |
| Cost | Higher | Lower |
| Suited for | Complex workflows | Budget-conscious high-volume work |
For example, a developer building a complex software project may prefer Sol for challenging coding tasks. Luna may be useful for someone who needs assistance in summarizing documents and drafting e-mails. Neither model is the right choice for every task. The right option depends upon the work and results you need.
GPT-6 Sol and Luna vs GPT-6 Astra
GPT-6 Astra remains OpenAI’s best model overall, according to the company. Sol and Luna offer affordable alternatives for users who want capable AI models at lower rates.
Here’s a simple way to choose the right model.
- GPT-6 Astra: For tasks that require OpenAI’s highest overall model capabilities are the priority.
- GPT-6 Sol: For demanding coding and professional tasks at a lower API cost than Astra.
- GPT-6 Luna: For everyday tasks and high-volume applications where lower API costs are a priority.
What Has Improved Over GPT-5.6?
Both models build on their predecessors, with improvements in coding, factual accuracy, and overall performance. In a specific evaluation of conversations where users flagged factual errors, GPT-6 Sol made about half as many mistakes as GPT-5.6 Sol, according to OpenAI. This result does not mean the model will make half as many mistakes in every situation.
OpenAI also reports better performance on coding and computer-based tasks, clearer communication, and fewer misleading claims about work completed. The company also fixed an image-encoding bug that affected how GPT-6 Sol and Luna processed images through the API and Codex. Together, these changes are intended to make the models more reliable when handling multi-step tasks. Developers can access GPT-5.6 Sol and GPT-5.6 Luna through the OpenAI API and Amazon Bedrock.
GPT-6 Sol and Luna Pricing
OpenAI has cut API costs by 50% compared with promotional pricing for GPT-5.6. The tables below compare the standard cost per million tokens for short and long contexts. Long-context rates apply when a request contains more than 272,000 input tokens.
GPT-6 Sol API pricing
| Token type | Short context | Long context |
| Input | $2.00 | $4.00 |
| Cached input | $0.20 | $0.40 |
| Cache writes | $2.50 | $5.00 |
| Output | $10.00 | $15.00 |
GPT-6 Luna API pricing
| Token type | Short context | Long context |
| Input | $0.10 | $0.20 |
| Cached input | $0.01 | $0.02 |
| Cache writes | $0.125 | $0.25 |
| Output | $0.50 | $0.75 |
OpenAI lets applications reuse previously processed input. Developers can save up to 90% on cached input tokens, particularly in applications that repeatedly use long prompts or conversation histories.
Unlike API pricing, ChatGPT access to GPT-6 Sol and Luna depends on your subscription plan, with Sol available on Plus, Pro, Business, and Enterprise plans, and Luna available on Free and Go plans.
Availability of GPT-6 Sol and Luna
OpenAI first released GPT-6 Sol and Luna through its API, ChatGPT Work, and Codex. On October 7, 2026, OpenAI expanded GPT-6 availability in ChatGPT, replacing GPT-5.6 Sol and Luna.
In regular ChatGPT, GPT-6 Sol powers the new model for Plus, Pro, Business, and Enterprise users, while Free and Go users receive GPT-6 Luna. The updated ChatGPT also includes Intelligent UI as part of the experience. Developers can access the models through the OpenAI API using the model IDs gpt-6-sol and gpt-6-luna. API charges depend on which model you use and its token consumption.
The October 7 update changed the models available in regular ChatGPT, not ChatGPT Work and Codex. The September versions of Sol and Luna remained available in ChatGPT Work and Codex, while the October versions powered the updated ChatGPT experience.
Suman has a background in Journalism and Mass Communication and works across content writing, editing, storytelling, and digital media. At Oratraex, she focuses on AI, guides, tools, and related digital topics, with an emphasis on creating content that is clear, practical, and easy to understand.
