High-throughput vector indexes now generally available and free

In December 2022, a new high-performance pod type, p2, became generally available, offering significant improvements in search speed and throughput compared to its predecessor, p1, by utilizing a graph-based index. These p2 pods, now included in the free Starter plan, provide users with the opportunity to experiment with high-throughput applications without any commitment, and are 50% cheaper due to optimized memory utilization. Compatible with dot product, cosine, and Euclidean distance metrics, p2 pods are recommended for applications requiring minimal latency and high throughput, such as video streaming recommendations and social media personalization, with pricing starting at $0.144 per hour as of December 1, 2022. Vertical scaling of p2 pods can enhance throughput capabilities, supporting over 1000 QPS with a single p2.x8 pod, making them ideal for performance-focused applications. Users are encouraged to test with their own data and consult documentation for optimal performance tuning, while those with different requirements might consider the s1 pod type.