RT nick | helius.dev (@nick_pennie)over 3 months we reduced p90 latency by 10x (7s to 700ms) for the enriched transaction history APIwithout touching its codesecret sauce coming soon https://t.co/3fIU7Dhnw4 ...

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]