Topic

Model inference cost

What model API calls cost for a unit of work, measured per task or per run rather than per token, including the gap between larger and smaller models.

Current clusters

build1 publisher

Handing a coding agent the frozen machine fixed 17 of 33 broken hackathon builds

Each frozen machine was rewound to its failing command, handed a coding agent, then graded by the harness rerunning that command itself. Seventeen of the 33 now build, and model time for the lot came to about $7.

Publishers:dev.to

Reality

Evidence55
Adoption12
Hype gap−10
Incentives30
Confidence48