API Performance Improvement - OpenAI's caching and API latency improvements have resulted in GPT-5 being significantly faster in Cursor [1] - P95 latency is approximately 2x faster than before, indicating a 50% reduction in latency [1]