Cost × Quality simulator

Move the sliders to see how model tier, tool calls, and iterations shift the cost and quality bands. The point is to find the knee in the curve — not to pick a winner.

Model tier

Input tokens / turn2000

Output tokens / turn800

Tool calls / run2

Agent iterations3

Surcharge per tool call (USD)$0.004

Cost / run

$0.0620

Input $0.0180 · Output $0.0360 · Tools $0.0080

Quality (illustrative)

91.0 / 100

Not benchmark data. Directional only.

Latency (est.)

3.2s

Includes tool call overhead.

Across tiers at these settings

Cost is real (based on current AI Gateway pricing for Claude Haiku/Sonnet/Opus). Quality is a simple illustrative heuristic — model tier + small tool boost − iteration penalty. Don't use it to pick a supplier.