14 agents · updated 2026-09-27. Higher index is better. Lower cost and time are better. Weigh a real prompt in the token counter or open the model directory.
Highlights
Coding Agent IndexHigher is betterTime per TaskLower is betterCost per TaskLower is better
Coding Agent Index vs. Cost per Task
Higher index and lower cost sit in the attractive quadrant.
Index vs costPareto line connects undominated agents
Token Usage per Task
Average input, cached input, and output tokens.
Output Cached input Input
TokensCache hit rate changes effective cost
Cost per Task
Average pay-per-token API cost. Lower is better.
CostUSD per task
Time per Task
Average agent wall time. Lower is better.
Execution timeActive runtimeIndex vs timeAttractive quadrant is fast and high-index