jev-codex-router GitHub avatar by 0xNatoshi

0xNatoshi /

Routing & Cost Optimization

jev-codex-router

Uses Jev to classify each Codex turn, then applies local rules to choose the model, reasoning effort, and speed mode.

License: MITLanguage: PythonLast push: 9/17/2026#llm-routing-cost#coding-agents

What Jev Does in this Architecture

Classifies task difficulty and reasoning needs before local policy selects a model configuration.

Key Architecture Benefit

Keeps routing policy and decision logs local for inspection and tuning.

Review & Benchmark Note: The roughly 60% saving is the author’s list-price replay estimate over 237 turns with fixed Token counts, excluding cross-model cache invalidation. It is not a reduction in a paid bill and was not independently retested.
Clone Repository
git clone https://github.com/0xNatoshi/jev-codex-router.git

Inspect on GitHub

Check out the upstream repository README, issues, and commit log.

0xNatoshi/jev-codex-router

Architecture Category

Routing & Cost Optimization

Complexity-based request routing to tier generative models and optimize token spend.

View all in this category →

Related Projects in Routing & Cost Optimization

jev-research repository icon by sherajdev

sherajdev

jev-research

1
Routing & Cost OptimizationMITTypeScript

A Jev and Herdr integration guide with a prototype for routing tasks to different Agents.

What Jev does here

Uses task and repository state to choose an executor and assess risk and dispatch readiness.

Provides a readable, editable example of multi-Agent task routing.

#llm-routing-cost#typed-decisions
2 evidence links
Inspect details
jev-demo repository icon by minghanminghan

minghanminghan

jev-demo

0
Routing & Cost OptimizationMITTypeScript

A customer-service routing demo that batches Jev questions before following the resulting route.

What Jev does here

Assesses classification levels, human-handoff intent and frustration, with low-confidence escalation.

Combines routing and human handoff while leaving response generation to the application.

#domain-workflows#classification-ranking
4 evidence links
Inspect details
Routing & Cost OptimizationMITTypeScript

A model-routing playground where Jev picks a candidate and users compare the resulting answers.

What Jev does here

Selects a model from the task and candidate profiles and records probabilities and runs.

Exports observations for checking routing choices against user preferences.

#llm-routing-cost#evaluation-benchmarks
2 evidence links
Inspect details