OpenAI Launches Sol Fast Mode and Cuts API Pricing
OpenAI has introduced Sol Fast mode for GPT-5.6 Sol, which runs up to 2.5x faster than the standard mode at roughly double the cost—about 25% cheaper per unit of speed. Simultaneously, the company reduced API pricing across its model lineup: Luna now costs $0.20/$1.20 per million input/output tokens, Terra drops to $2/$12, and Sol to $5/$30.
With these changes, Luna is now 25x cheaper than Sol, targeting high-volume tasks like classification, data processing, support bots, and simple agents. Sol Fast is aimed at latency-sensitive use cases: interactive products, complex agents, and real-time workflows. OpenAI is pushing on both fronts—cheap inference for scale and accelerated flagship models for performance.
Related: OpenAI Previews GPT-5.6 Sol