2 essays
Speed Tests
Measured first-token latency, tokens per second, total response time, consistency, and concurrent performance of fast AI models.
01
Does GLM 5.3 Fast Slow Down Under Concurrent Requests?A transparent method for testing whether GLM 5.3 Fast holds its speed when several request...
Speed Tests
Oct 18, 2026
→
02
Ten Repeated Runs: Measuring Variance Instead of Cherry-Picking the Best ResultWhy a single speed number is not enough: how Speed Index uses ten repeated runs to measure...
Speed Tests
Oct 15, 2026
→