[ACL 2026] Official implementation of MTRouter, a cost-aware multi-turn LLM routing framework accepted to ACL 2026 Main Conference.
-
Updated
Jun 16, 2026 - Python
[ACL 2026] Official implementation of MTRouter, a cost-aware multi-turn LLM routing framework accepted to ACL 2026 Main Conference.
ORBIT: A Unified Toolbox for LLM Routing. Unified LLM/model routing, cost-aware inference, multimodal benchmarks, and reproducible evaluation. Accepted by Frontiers of Computer Science (2026).
NeurIPS 2025 paper "MESS+: Dynamically Learned Inference-Time LLM Routing in Model Zoos with Service Level Guarantees"
Unified cost-aware router: model x reasoning-depth on one Thompson bandit (factored policy), bi-temporal forget-on-drift statistics memory + auto-drift proxy, GRPO-internalized Qwen3-4B depth policy.
Universal, zero-runtime-dependency LLM router for Node, edge, and browser. Routes every prompt to the right model by cost, latency, and measured quality. Native streaming, tool calling, retries, circuit breaking, and budget guards. Drop-in adapters for OpenAI, Anthropic, Gemini, any HTTP endpoint, and the Vercel AI SDK.
Cost-aware routing of factuality verifiers from cheap pre-call features: reproducibility release
Workflow registry and cost-aware routing policy layer for local coding agents.
Learned deliberation routing (System III Configurator) from 'Critique of Agent Model' (arXiv:2606.23991): a cost-aware bandit over reasoning depths (direct/cot/plan+verify), live on Vi-GSM8K, with a GRPO path for Qwen3-4B.
RL agents across LangChain & LangGraph, 50+ MCP registry tools, PGVector stores, deployment w Dynamo and inference with vLLM & SGLang
To associate your repository with the cost-aware-routing topic, visit your repo's landing page and select "manage topics."