Extension Benchmark Study

41.9% Lower Aggregate AI Model Cost

Across seven controlled extension benchmarks, candidate configurations used $13.1193 versus $22.5774 for their baselines. Each extension was evaluated separately; this cost-weighted portfolio result does not measure all extensions enabled together.

September 20267 comparable extensionsIndividual cost reductions: 14.5%–66.0%

Extension Results

Quality is the candidate aggregate pass rate. Lower is better for cost, token, and duration measures.

Candidate quality, cost change, and secondary outcomes for seven comparable Felan extensions
Extension	Quality vs baseline	Cost vs baseline	Secondary result
Subagents	100%	23.7% lower*	—
MarkItDown	100%	31.0% lower	13.8% fewerPrompt tokens
Concise Output	100%	14.5% lower	16.4% fewerOutput tokens
Prewalk	100%	66.0% lower	—
RTK	83.3%	26.6% lower	40.6% fewerPrompt tokens
Session Compaction	100%	45.2% lower	52.1% fewerPrompt tokens
Codebase Memory	100%	27.8% lower	7.8% shorterAgent-step duration