fewer fresh tokens
Median fresh input tokens across the benchmark suite.
We built MOAH to test a simple question: can an agent keep access to more capabilities without paying the context cost of exposing everything at once?
| Metric | Pi | MOAH | OpenCode |
|---|---|---|---|
| Fresh tokens | 1.00× | 1.44× | 3.64× |
| Processed | 1.00× | 1.30× | 6.96× |
| Success | 17/19 | 17/19 | 14/19 |
| Median E2E | 4.9s | 6.1s | 11.2s |
Pi starts with a deliberately minimal tool surface. MOAH explores whether a larger capability library can remain available while keeping the active working set small.
Median fresh input tokens across the benchmark suite.
Median processed tokens across the benchmark suite.