Observed arrival · 2026-11-18
Merit Order routes LLM requests by the cost of getting the work done
This pre-production, self-hosted gateway chooses a model for each request based on its quality threshold and expected cost.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Developers routing requests across multiple LLMs
- Worth noticing
- The simulator compares 2,000 mocked tasks and explicitly says its results are not measured on live traffic.
Field notes
The routing explanation estimates each model’s chance of finishing a task and includes the cost of failure, another attempt, and rebuilding the prompt cache elsewhere. The page also shows a threshold-based alternative and worked examples for other mechanisms, including prompt caching. Its comparison figures are tied to a simulator or list-price arithmetic, not reported live usage.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
One card from the complete issue