New paper tests 5 compaction strategies: long-horizon agent benchmarks are surprisingly compaction-friendly
maksym_andr · x · 2026-10-31
New paper "Benchmarks Are Surprisingly Compaction Friendly" presents the first systematic evaluation of context compaction on long-horizon agentic benchmarks. Agents routinely exceed their context windows and compact to keep going — nearly all frontier agents do this — yet no one had evaluated compaction on real long-horizon benchmarks. The team tested 5 compaction strategies and found the benchmarks tolerate compaction surprisingly well.
More from coding & agent
- krishnan on autonomous lab agents: criticism changed the next search, not just verdicts — krishnan · 2026-10-31
- Testing agents with explicit rejection records: a cheaper path to verified troubleshooting answers — krishnan · 2026-10-31
- krishnan proposes logging ruled-out troubleshooting hypotheses for future reopening — krishnan · 2026-10-31
- Multi-agent AI lab autonomously researches quantum spins in diamond, revises drifting experiments — krishnan · 2026-10-31
- PI agent rejected 75 proposals in 30 hours; manager banned recurring bad ideas — krishnan · 2026-10-31
- Next 9 proposals explored 7 new directions: rejection became information — krishnan · 2026-10-31