Simon Willison • 7/30/2026

Advancing the price-performance frontier with GPT‑5.6

OpenAI announced significant price reductions for GPT-5.6 models: Terra drops 20% and Luna drops 80%. The post explains how GPT-5.6 Sol was used to optimize load balancing and the model's forward pass, including rewriting production kernels in Triton and Gluon, reducing serving costs by 20%. With Luna at $0.20/M input and $1.20/M output, it undercuts Google's Gemini 3.1 Flash-Lite and Anthropic's Claude Haiku 4.5. The author switched their demo site from Gemini to Luna, highlighting the competitive shift in AI pricing.

0 comments

#Openai #Inference Optimization #Gpu Programming