What is Pareto Code Router?
OpenRouter's coding router that sends each request to the cheapest model above your chosen coding score.
Who is Pareto Code Router best for?
Pareto Code Router is best for developers who want a strong coding model behind one OpenRouter model ID.
Coding agents on a budget
A lower min_coding_score routes to cheaper models that still beat the median of Artificial Analysis's coding field.
Always using a top coder
Leaving the score out, or setting 0.66 or higher, routes to the top tier of current coding models.
Fast responses
The :nitro variant picks the fastest model in your tier by throughput instead of the cheapest.
Long coding sessions
Session stickiness keeps a conversation on the same model and provider, which helps prompt-cache hits.
How much does Pareto Code Router cost?
Pareto Code Router adds no fee of its own: each request is billed at the OpenRouter price of the coding model it routes to.
The router has no fixed price. Each request is billed at the rate of the model that serves it, so cost shifts with the tier you choose and with the shortlist as new models arrive. Plans and current prices are on openrouter.ai.
How do readers rate Pareto Code Router?
No one has rated Pareto Code Router yet. Be the first.
Sign in to rate. One rating per account.
What are the best Pareto Code Router alternatives?
One tool on Recaply does Pareto Code Router's job.
Questions about Pareto Code Router
What is Pareto Code Router?
OpenRouter's coding router that sends each request to the cheapest model above your chosen coding score.
Who is Pareto Code Router for?
Pareto Code Router is best for developers who want a strong coding model behind one OpenRouter model ID.
Is Pareto Code Router free?
No. Pareto Code Router is a paid tool.
What are the best Pareto Code Router alternatives?
The closest alternatives on Recaply are Cursor Router.
Where can I try Pareto Code Router?
On its official website, openrouter.ai.
How does Pareto Code Router pick a model?
It maps your min_coding_score to a high, medium or low tier based on Artificial Analysis coding percentiles, then picks the cheapest available model in that tier, with up to two same-tier fallbacks for provider errors. The response names the model that answered.







