Looplane uses a static model-role catalog and retries or falls back only after retryable provider errors. Cache data is a provider hint plus trace, while cost is a static-table estimate; neither is live routing intelligence or a bill.
LiteLLM is not a model provider. It is a Python SDK and self-hosted proxy that normalizes 100+ LLM APIs, then centralizes routing, fallbacks, virtual keys, budgets, and observability at the gateway layer.
OpenRouter exposes many models and inference endpoints through an OpenAI-compatible API, with provider ordering, failover, BYOK, and zero-data-retention controls in one routing policy.
Portkey sits between applications and model providers: one OpenAI-compatible endpoint adds routing, fallbacks, request logs, budgets, and guardrails, with an open-source gateway available for self-hosting.
Spin up a local OpenAI-compatible endpoint at localhost:20128 that automatically routes requests from Claude Code / Cursor / Cline / Codex / Copilot through a Subscription → Cheap → Free 3-tier fallback to 40+ providers. Built-in RTK compresses tool_result (saving 20–40% input tokens), Caveman mode compresses output, OAuth auto-refresh, multi-account round-robin — install with npm install -g 9router and two commands.