GPT-5 Mini

GPT-5 Mini: The cost-optimized workhorse model, engineered for high-volume, low-latency applications that demand strong reasoning at scale.

This is the system's default workhorse: a highly efficient model balancing capability, speed, and cost. GPT-5 Mini is specifically designed for cost-optimized reasoning and high-throughput chat applications (e.g., customer service, data extraction). It maintains strong performance for well-defined tasks, offering a median throughput of approximately 170 tokens per second (TPS)—significantly faster than its predecessor, GPT-4.1. Developers benefit from a massive pricing advantage: processing four million input tokens on Mini costs the same as just one million on the flagship GPT-5. Use Mini as your primary engine for production-level tasks like code generation, structured data parsing, and conversational AI, reserving the full GPT-5 for only the most complex, multi-step agentic workflows.

https://openai.com/gpt-5