Third-party cyber evaluations involving OpenAI models
openai.com·13h ago
Data Science
Simon Willison's Blog·Jun 26·5 min read
Quoting OpenAI
TL;DR
OpenAI identified a need for more accessible and cost-effective AI models. They developed the GPT-5.6 series, which includes Sol, Terra, and Luna, each tailored for different use cases.
✦ Why It Matters
Engineers can leverage these new models to optimize costs while maintaining high performance in AI applications.
Key Takeaways
- Explore the cost benefits of using Terra and Luna for your applications.
- Implement prompt caching to improve response times and reduce costs.
- Stay updated on the general availability of GPT-5.6 models.
- Consider participating in the limited preview if eligible for early access.
Full Summary
The GPT-5.6 series introduces three models: Sol, a flagship model; Terra, a cost-effective option for everyday tasks; and Luna, an affordable and fast model. Terra offers performance comparable to GPT-5.5 at half the cost, while Luna provides strong capabilities at the lowest price point.
Pricing is structured per million tokens, with Sol costing $5 input and $30 output, Terra at $2.50 input and $15 output, and Luna at $1 input and $6 output. Additionally, GPT-5.6 enhances prompt caching with features like explicit cache breakpoints and a minimum cache life of 30 minutes.
This series aims to provide broader access to advanced AI capabilities, starting with a limited preview for select partners before a wider release.