Llm Routing Python Packages

xrouter-llm

A prompt-aware LLM router that predicts which models can complete each request, then selects the cheapest capable one: 53.2% lower cost and +1.9 pts completion on our tested dataset.

orchestrator-mcp-server

Use Codex from Claude Code, or Claude Code from Codex, through the subscriptions you already have. Multi-turn consultations between agent CLIs with no provider API key, a validated response envelope, and a local dashboard of every consultation.

ratel-ai

Context engineering for AI agents. ~80% fewer tokens. Fix tool overload. Skills and memory with in-process BM25 and semantic retrieval. Progressive Disclosure. No vector DB.

calybris

Deterministic proof-carrying decision core with replay verification, WAL, and fixed-point budget proofs

cafitac-hermit-agent

Quiet MCP executor for Claude Code and Codex. Offload edits, tests, refactors, and commits to cheaper local or flat-rate models.

kalibr

Stop overpaying to run your agents. Kalibr routes every request to lower-cost model and tool paths without degrading performance.

jev-style

Small, calibrated decision models on your own machine: systemone-compatible local server, 6 agent skills, Claude Code guard, MCP tools. Weights on Hugging Face.

a3m-router

⚡ Adaptive multi-model LLM router — 80+ providers, Jev System One single-pass routing (model=jev-auto), pheromone-trail failover, parallel ensemble merge. npm: adaptive-memory-multi-model-router

ratel-ai-telemetry

Context engineering for AI agents. ~80% fewer tokens. Fix tool overload. Skills and memory with in-process BM25 and semantic retrieval. Progressive Disclosure. No vector DB.

cre-router

Cascaded Framework for Cost-Aware LLM Serving

evalt

Budget-bounded LLM routing that finds the cheapest model and prompt meeting your accuracy target.

thinkless

The decision plane for AI agents. Rules and small calibrated models make the routine decisions, the LLM is reserved for the steps that need thought, and every step is traced.

claude-router

claude-router is a local prompt router that picks the right Claude model tier and prepends the right scaffold using local embeddings before you call the API. A deterministic routing layer for eval, research, content, and review prompts that helps teams stop overspending on Sonnet and Opus when Haiku plus structure is enough.

kry-attest

Tamper-evident, stranger-verifiable receipts for LLM cost-savings — anyone can recompute your caching/routing savings math offline, no trust in your dashboard required. Pure stdlib, zero-dependency.

spekoai

Official Speko Python SDK (spekoai) - one API for every voice AI provider: STT, TTS, and LLM routing + failover, sync + async

claude-wayfinder

Deterministic, auditable agent/skill dispatch for Claude Code. A typed seven-decision scoring kernel runs after the router has read the conversation — post-cognitive routing instead of prose-scanning. Auto-generates its catalog from skill sidecars and agent frontmatter.

agentplane

Define project policy once, render it into every agent harness, route work across providers, and get auditable results.

agentweave-router

Pre-inference tool and agent routing for MCP, A2A and tool-rich LLM systems. Reduce candidate tools before model inference with policy-aware routing and recovery.

tazuna

Define project policy once, render it into every agent harness, route work across providers, and get auditable results.

blackdrome-noma

Noma: an open-weight decision model for AI agents. Calibrated decisions in milliseconds, self-hosted, with a Jev-compatible API.

baar-core

Budget-Aware Agentic Routing (BAAR) — Intelligent LLM model selection with a zero-call financial kill-switch. Save 90% on costs without losing accuracy.

langchain-skill-router

Per-turn skill selection for LangChain and deepagents agents: a fast judge picks the few skills a turn needs, so a catalog of hundreds stays out of the prompt.

verdictml

Small, fast, honest decision models. Open alternative to Jev: zero-shot, fit on your labels in seconds, calibrated with a coverage guarantee, Jev wire-compatible.

llm-pqr

Provider-neutral LLM model selection using measured quality, latency, cost, capabilities, and hard privacy constraints.