Type Alias: CostPreset

CostPreset = "quality" | "economy"

Defined in: apps/paracosm/src/cli/sim-config.ts:380

Cost-vs-quality preset for runSimulation model routing.

• 'quality' (default): DEFAULT_MODELS — departments on flagship (gpt-5.4 / claude-sonnet-4-6) for reliable tool forging, other roles on mid/cheap tier. Per-run spend ~$1-3 on OpenAI, ~$3-7 on Anthropic at 6 turns × 5 depts × 100 agents.

• 'economy': DEMO_MODELS — departments on gpt-4o (half the cost of flagship, adequate for forge shape compliance on most scenarios), everything else on gpt-5.4-nano / Haiku. Per-run spend drops to ~$0.20-0.60 on OpenAI. Use for quick iteration; forge approval rate drops ~10-20pp vs quality.

Explicit models overrides always win over the preset.