What it does
A command line tool that picks which model tier a coding task should run on and then records whether the result was actually good, so later decisions use those verified outcomes instead of a fixed rule. It installs from a prebuilt binary, so no Rust toolchain is needed, and it keeps its policy file and history local to the machine. The project is open source on GitHub.
Teams route coding tasks between model tiers by hand, and the choice never learns from which attempts actually worked.
- Stage
- Live
- Made for
- Developers running coding agents, Platform engineers managing model spend, Teams comparing model tiers
- Platforms
- Web
- Company
- Jevia
- Website
- jevia.dev ↗
Timeline
- Jevia released as an outcome-based model router for coding agents
A CLI that routes tasks to a model tier and improves the routing from verified outcomes.
First release · Took effect · Proposed by @Jeetaksh25
Sources
Contributor credit
This profile adapts text by @Jeetaksh25 from profile PR #14, licensed under CC BY 4.0. Its factual claims were reviewed separately.
Publication record
- Release
sha256:d10780775c52fe0c0c69ed41ff11a32b7312483d04a4a984aee6905ac692db8f- Revision
sha256:57746b271804653090bf96412272dfe7ba96dff8518c96ec763a65348b93e755
Every statement here carries its date and stays true as of that date.