[Alex Perry](https://mashable.com/author/aperry)
[yelix.bsky.social](http://yelix.bsky.social).
[Read Full Bio](https://mashable.com/author/aperry)
Not to be outdone by Anthropic, OpenAI is launching a couple of new ChatGPT models on Tuesday.
Mere hours after the launch of Claude Opus 5.5, GPT-6 Sol and Luna became available to users, per an OpenAI blog post. These new models are said to provide better accuracy at a more cost-efficient value than prior models. They can also be thought of as less-intensive versions of GPT-6 Astra, which launched earlier in September. OpenAI says Sol and Luna are better for work at different scales than what you might use Astra for.
"We trained GPT‑6 Sol and Luna with similar methods as GPT‑6 Astra, bringing the advances behind Astra’s state-of-the-art performance in professional work, factuality, coding, computer use, and alignment to faster, more affordable models," OpenAI's blog post said.
[Terms of Use](https://www.ziffdavis.com/terms-of-use)and
[Privacy Policy](https://www.ziffdavis.com/ztg-privacy-policy).
You May Also Like
[This Tweet is currently unavailable. It might be or has been removed.](https://twitter.com/ChatGPT/status/2102462020339613966)
If you want to try these new models right now, they're available in Work and Codex for Plus, Pro, Enterprise, Business, and Edu customers. Go and Free users can access Luna in the ChatGPT desktop app, but be warned: neither model is available in basic Chat just yet. Here are the prices for the new models, per 1 million tokens:
- **GPT-6 Sol:** $2 input, $10 output
- **GPT-6 Luna:** $0.10 input, $0.50 output
Those prices represent a 50 percent decrease in cost compared to GPT 5.6 Sol and Luna. Meanwhile, OpenAI says GPT-6 Sol and Luna outperform those older models to the tune of making half as many factual errors as before. It's worth noting that the 5.6 variants launched just a few months ago, so that means (if OpenAI is telling the truth) the company was able to cut down on errors by half in just one financial quarter.
If you want hard numbers to back that up, OpenAI says GPT-6 Luna beat the 5.6. release by 5.4 percent, using AutomationBench as a measuring stick, and it did so while costing 58 percent less per task. There are a handful of other benchmarks used in the blog post, and the new GPT-6 models come away looking better than their predecessors in all of them, naturally. Disclosure: Ziff Davis, Mashable’s parent company, in April 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.
Topics
[Artificial Intelligence](https://mashable.com/category/artificial-intelligence)
[ChatGPT](https://mashable.com/category/chatgpt)
[OpenAI](https://mashable.com/category/openai)
Alex Perry is a tech reporter at Mashable who primarily covers video games and consumer tech. Alex has spent most of the last decade reviewing games, smartphones, headphones, and laptops, and he doesn’t plan on stopping anytime soon. He is also a Pisces, a cat lover, and a Kansas City sports fan. Alex can be found on Bluesky at yelix.bsky.social.