Choosing a strategy

The public v1 service contract supports two strategies: greedy and alns.

Pick by time budget and by how much search depth you need for the job.

TL;DR

Strategy	Time budget	Quality profile	Best for
greedy	Low latency	Fast baseline	Live previews, instant quotes, synchronous UX
alns	Higher latency	Deeper search, better layouts on hard jobs	Waste-sensitive production runs, batch optimization

Greedy

Use greedy when latency matters more than squeezing out the last few percent of utilization.

{ "strategy": "greedy" }

• Fastest path through the public API.
• Good default for interactive workflows.
• Minimal configuration surface.

ALNS

Use alns when you can afford a deeper search budget and want better material efficiency.

{
  "strategy": "alns",
  "alnsConfig": {
    "maxIterations": 250,
    "timeLimitMs": 250,
    "numThreads": 1,
    "seed": 42
  }
}

• Adds an explicit optimization budget.
• Supports seeded runs for reproducibility.
• Can return top-level alnsStats in the response.

Practical rule

• Start with greedy when the caller is waiting on screen.
• Move to alns when material savings are worth extra solve time.
• Use the same input contract for both; only strategy and alnsConfig change.