v1.2.0 v3.8.44

Name: OmniRoute
Author: cheaper-inference

high risk

cheaper-inference/OmniRouteJul 4, 2026 → Jul 4, 20260 days between releases

What's New

Adds quota throttling, new providers, and auto-combo controls; major version jump.

Breaking Changes

Configuration schema likely changed due to new relay backend and reverse-proxy basePath options.
Core routing logic shifted from council-centric to a generic proxy/relay model.

New Features

Per-request Auto-Combo controls via headers (`X-OmniRoute-Mode`, `X-OmniRoute-Budget`)
New providers: Kenari, Mistral OCR, Discovery tool, Bifrost, Mux
Selectable relay backend (TypeScript, Bifrost, or auto)
xAI usage tracking
CodeWhale CLI tool

Improvements

Throttle upstream quota fetches to prevent token revocation
Improved i18n auto-detection
Quota collapse/sort and bulk-sync spacing

Bug Fixes

Added regression guard tests for quota fetch throttling

Upgrade Notes

Configure new headers `X-OmniRoute-Mode` and `X-OmniRoute-Budget` for per-request control. Set `OMNIROUTE_QUOTA_FETCH_MIN_INTERVAL_MS` (default 250ms) to prevent token revocation. Expect configuration schema changes due to the new relay backend options.

The Two Releases

v1.2.0

Jul 4, 2026· 4 months ago

This release enhances the council with plugin marketplace support, advanced confidence-weighted voting mechanics, and project-level configuration overrides.

Full release notes

v3.8.44

Jul 4, 2026· 4 months ago

Focuses on quota fetching, auto-combo controls, and new providers like Kenari and Mistral OCR.

Full release notes