Artificial Intelligence
OpenAI unveils cheaper GPT-6 Sol and Luna
OpenAI launched GPT-6 Sol and GPT-6 Luna, said they used methods similar to Astra, and priced them at roughly half the cost of earlier versions while widening availability across its products.
- OpenAI introduced GPT-6 Sol and GPT-6 Luna on Tuesday.
- The company said the models were trained with methods similar to Astra and were aimed at factuality, coding, computer use and alignment.
- GPT-6 Sol was priced at $2 per million input tokens and $10 per million output tokens.
- GPT-6 Luna was priced at 10 cents per million input tokens and 50 cents per million output tokens.
- OpenAI said GPT-6 Sol outperformed previous OpenAI models and Anthropic’s Claude Opus 5 on AutomationBench, Agents’ Last Exam and DeepSWE v1.1.
- The models were available in ChatGPT Work and Codex for Plus, Pro, Business and Enterprise users, while Free and Go users could access GPT-6 Luna in the desktop app.
On Tuesday, OpenAI introduced GPT-6 Sol and GPT-6 Luna. The company said both models had been trained using methods similar to Astra and were intended to improve factuality, coding, computer use and alignment.
OpenAI set GPT-6 Sol’s price at $2 per million input tokens and $10 per million output tokens. GPT-6 Luna was priced at 10 cents per million input tokens and 50 cents per million output tokens, which the article said was about half the cost of the prior versions.
OpenAI said GPT-6 Sol beat earlier OpenAI models and Anthropic’s Claude Opus 5 on several tests, including AutomationBench, Agents’ Last Exam and DeepSWE v1.1. It also said the two models used Astra’s clearer style, with shorter answers and fewer filler details, and that their alignment had improved through lower rates of warning circumvention, coding deception and unauthorised agent use.
The models were available in ChatGPT Work and Codex for Plus, Pro, Business and Enterprise users. Free and Go users could use GPT-6 Luna in the desktop app, and the models were not yet available in traditional chat.
Named in this story
Companies
- OpenAI
- released the new models and set their pricing
- Anthropic
- rival company named in the benchmark comparisons
Products and systems
- GPT-6 Sol
- new model priced at $2 and $10 per million tokens
- GPT-6 Luna
- new model priced at 10 cents and 50 cents per million tokens
- Astra
- earlier model whose training methods were cited as a basis
- Claude Opus 5
- Anthropic model used in OpenAI’s benchmark comparisons
- 5.6 Sol
- previous generation used for the pricing comparison
- ChatGPT Work
- one of the products where the models were made available
- Codex
- another product where the models were made available
How the source tells it
The article read like a product launch, with OpenAI’s own claims and rival comparisons carrying the narrative.
- vendor boosterism the launch was built around OpenAI’s own capability and pricing claims
- competitive framing rival comparisons were used to measure the new models against Anthropic and earlier OpenAI versions