Advancing the price-performance frontier with GPT‑5.6
Advancing the price-performance frontier with GPT‑5.6 Huge price drop from OpenAI today: GPT-5.6 Terra got a 20% reduction, and GPT-5.6 Luna got a massive 80% drop. OpenAI credit 5.6 Sol with enabling this: in How GPT‑5.6 fuses frontier intelligence with frontier efficiency they describe using 5.6 Sol to optimize load balancing, and more impressively to optimize inference itself: We also used…
OpenAI has slashed the prices of its latest AI model, GPT-5.6, in a significant move to advance the price-performance frontier. The company announced a 20% reduction in the price of GPT-5.6 Terra and an even more substantial 80% drop in the price of GPT-5.6 Luna. This move puts GPT-5.6 Luna at a fraction of the cost of other competing models, making it significantly cheaper than Google's Gemini 3.1 Flash-Lite and Anthropic's Claude Haiku 4.5.
In a detailed technical paper, OpenAI explained how they achieved these price reductions. They leveraged GPT-5.6 Sol to optimize various aspects of the model, including load balancing, inference, and even the production kernels. By finding work that could be precomputed, avoided, or parallelized, GPT-5.6 Sol was able to reduce excess memory movement and synchronization, ensuring that GPUs were utilized efficiently.
The optimizations made by GPT-5.6 Sol led to a 20% reduction in end-to-end serving costs. This development marks a major shift in the landscape of lower-priced AI models, with Luna now offering a level of affordability previously unseen. As a result, Simon Willison, a developer who previously used Google's Gemini 3.1 Flash-Lite for his agent.datasette.io demo site, has switched over to Luna, highlighting the significant impact this price drop has had on the AI landscape.
Written by urgent.news from Simon Willison's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
