Capability per dollar
Benchmarks & value for money
Independent benchmark scores joined to our own verified price data — so the leaderboard answers the real question: which model buys the most capability per dollar?
01 Price–performance frontier
log price · scoreNot enough models with both a tracked price and a score to plot the frontier yet.
02 leaderboard
higher is betterNo observations for this benchmark yet. Run bin/rails benchmarks:run.
Score / $1M is the benchmark score divided by the blended input+output price per 1M tokens, so it ranks capability per dollar rather than raw capability. Rows without a tracked price sort last.
Scores © their sources, used with attribution: Epoch AI benchmark data (CC BY 4.0) and LMArena (CC BY 4.0). Prices are PriceDrift's own verified captures.