Introducing GPT-5.4 mini and nano

Build a high-performance inference server optimized specifically for 'nano' parameter models, targeting sub-50ms latency for sub-agent orchestration.

Suggested repo: nano-serve

"Deploy your micro-models at sub-millisecond speeds."

Estimated effort: 200h