Benchmark Results

Config: default (summarization memory) · Model: Claude Sonnet · 6 tasks · March 2026

Steps per task

Token usage

Metrics overview

Input vs output tokens

Task results

TaskStatusTestsStepsInput tokensOutput tokens

Agent trajectory