Live Benchmarks

Prembly Benchmark

Performance results of AI coding models on Prembly tasks, measuring success rate and execution time with high precision.

View on GitHubTotal tasks: 12Last run: 4/22/2026

Model Performance

ModelPassedAvg DurationSuccess Rate
#1
gemini-3-flashNEW
8294.8s
67%
#2
glm-4.7
7473.2s
58%
#3
gemini-3.1-pro
7291.4s
58%
#4
claude-4-6-sonnet
6326.7s
50%