From OpenAI · · 2m
OpenAI Launches GPT-6 Sol and Luna for Cost-Efficient AI
OpenAI releases GPT-6 Sol and Luna to cut AI costs while maintaining high performance for everyday tasks.
In brief
OpenAI releases GPT-6 Sol and Luna to cut AI costs while maintaining high performance for everyday tasks. OpenAI introduces GPT-6 Sol and Luna models that reduce API costs by 50% compared to GPT-5.6 while improving performance in coding, factuality, and computer use. Originally reported by OpenAI.
New GPT-6 Models Launch
OpenAI introduces GPT-6 Sol and Luna to expand the GPT-6 family with cost-efficient intelligence for everyday work tasks.
Cost Efficiency Focus
GPT-6 Sol and Luna reduce API prices by 50% compared to GPT-5.6 promotional pricing for input and output tokens.
Professional Work Performance
GPT-6 Sol outperforms Claude Opus 5 at 9% of its cost per task on AutomationBench for business workflows.
Factuality Improvements
GPT-6 Sol makes half as many factual errors as its predecessor, approaching Astra-level reliability at lower cost.
Coding Capabilities
GPT-6 Sol improves coding performance on FrontierCode and DeepSWE tests while reducing costs by up to 80% per task.
Computer Use Efficiency
GPT-6 Sol matches Claude Opus 5 at 80% lower cost per task on OSWorld 2.0 for computer workflows.
Collaboration Style
GPT-6 Sol provides clearer, more concise responses with fewer jargon and low-value details in technical conversations.
Prompt Caching Improvements
OpenAI enhances prompt caching to reduce fresh processing by 50% across billions of requests for developers.
Alignment Progress
GPT-6 Sol and Luna show improved alignment over GPT-5.6 with lower rates of misleading claims about coding work.
Availability Details
GPT-6 Sol and Luna available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users starting today.
Model Pricing Comparison
GPT-6 Sol costs $2 per million tokens versus $4 for GPT-5.6, and GPT-6 Luna costs $0.10 versus $0.20 for GPT-5.6.
Real-World Example
GPT-6 Sol reshapes website layouts without needing React, using native page transitions for smoother performance.
Gradual Rollout Plan
OpenAI rolls out GPT-6 Sol and Luna gradually in ChatGPT Work and Codex throughout the day to ensure stable service.
Next Steps
Developers can monitor caching performance and optimize their applications using the Prompt Caching Dashboard and diagnostics tools.
In short
OpenAI introduces GPT-6 Sol and Luna models that reduce API costs by 50% compared to GPT-5.6 while improving performance in coding, factuality, and computer use.
