LiteLLM Adds Modality- and Context-Window-Based Auto-Routing

LiteLLM (YC W23) · 2026-09-01

Key Info

LiteLLM now supports automatic routing that classifies each input by modality and context window, then selects the best-suited model and reroutes to capable models when needed.

Highlights

Inputs are classified by modality (e.g., text, image, audio) to pick the best model for that type.
Context-window size is factored into routing, so requests are rerouted to models that can handle them.
Aimed at simplifying multi-model workflows and improving reliability across different LLM tasks.