OpenAI reduces GPT-5.6 model pricing

OpenAI has announced significant price reductions for its GPT-5.6 model family. The Luna model saw an 80% price cut, bringing costs to $0.20 per million input tokens and $1.20 per million output tokens. The larger Terra model received a 20% reduction, resulting in costs of $2 per million input tokens and $12 per million output tokens.

The company attributed these reductions to internal optimizations driven by the GPT-5.6 Soul model, which utilized kernel rewrites and speculative decoding to achieve efficiency gains of 20-35% in inference workloads.

These pricing changes occur amid increasing competition from Anthropic and various Chinese AI laboratories. For comparison, Moonshot AI’s Kimi K3, an open-weight model with 2.8 trillion parameters, is priced at $3 per million input tokens and $15 per million output tokens. The shift reflects a broader trend where enterprise customers utilize model routers to direct tasks to the most cost-effective model available.