GPT-5.6 API Prices Slashed: Entry Model Down 80%, Flagship Gets 2.5x Speed Boost

量子位 · wechat · 2026-07-31

OpenAI has announced massive price cuts and speed boosts for its GPT-5.6 API series, intensifying the LLM price war.

OpenAI attributed the cost reductions to GPT-5.6 Sol optimizing its own production systems (e.g., rewriting GPU kernels and improving speculative decoding), which lowered end-to-end serving costs by 20%. This move significantly lowers the barrier for running continuous Agent tasks, initiating a cost-reduction flywheel driven by the model itself.

Related event: OpenAI Launches GPT-5.6: Luna Price Drops 80%, Sol Self-Optimizes(63 posts)→

Original post →

More from Models

Models channel →