Propose a part

Datasheet · A prompt line

gm ser

gm. stay based, ser.

Folklore, as cited: crypto Twitter

Hurts on DeFi maths, Haiku 5.5 with thinking off (−3.7 points); changes nothing that matters on DeFi maths, Haiku 5.5 with thinking on; changes nothing that matters on Agent safety, Haiku 5.5 with thinking off.

DeFi maths, Haiku 5.5 with thinking off

Pooled across 2 notebooks: −3.7 points, 95% interval −6.6 to −0.8; measured once.

  1. UNRESOLVEDNotebook #017−2.3 points (−6.5 to +1.9), 432 pairs, cost ×0.95signed verdict 107
  2. INHIBITSNotebook #018−5.0 points (−9.0 to −0.9), 432 pairs, cost ×0.97signed verdict 112

DeFi maths, Haiku 5.5 with thinking on

Pooled across 1 notebook: −0.9 points, 95% interval −4.0 to +2.2; measured once.

  1. INERTNotebook #021−0.9 points (−4.0 to +2.2), 432 pairs, cost ×0.98signed verdict 133

Agent safety, Haiku 5.5 with thinking off

Pooled across 1 notebook: +0.1 points, 95% interval −1.4 to +1.7; measured once.

  1. INERTNotebook #012+0.1 points (−1.4 to +1.7), 432 pairs, cost ×0.99signed verdict 70

Measure it in a setting it has not been measured in: propose it. As JSON for an agent: /v1/parts/6db2f4261266fa15. Every part: the atlas.