Provisional tally 72.7%… Alma, ahead of Claude Code with lower success cost
Original headline잠정 집계 72.7%…알마, 클로드 코드 앞서며 성공 비용도 낮아
Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.
Independent developer Yetone's AI agent Alma posted a 72.7% task pass rate on a snapshot of the FrontierHarness v1.1 leaderboard, edging past Claude Code at 70.0%, while Codex CLI led at 77.9% and Pi reached 72.8%. Average cost per successful task was $0.80 for Alma versus $0.94 for Claude Code, a $0.14 gap measured on tokens consumed during task execution. The results matter because they suggest a solo-built agent can match or beat established coding tools on both pass rate and cost efficiency. The figures are provisional, and the next thing to watch is whether Alma holds its position as the leaderboard fills out with fuller evaluations.
Muhabbit's AI wrote this summary in its own words from TokenPost's report and translated it from Korean. It can get details wrong, so check the source.
Named in this story
Alma, Claude Code, Yetone, FrontierHarness, Codex CLI