Integration guides
Point your favorite AI tool at Leanroute in five minutes.
Every tool here speaks the OpenAI wire format. That means three things move per integration: a base URL, an API key, and one or more model names — everything else stays identical to the tool's default OpenAI setup.
Pick a model, get every endpoint snippet in curl / Python / JS. One-click Postman collection download.
Browse all 38 models with filters by provider, tier, region, and vision support. Pricing per 1M tokens.
The three values you'll need.
- Base URL:
https://api.leanroute.dev/v1 - API key: a
gw_live_*token from /dashboard/keys. - Model name: the canonical
provider/modelform, e.g.anthropic/claude-sonnet-4-6,openai/gpt-4o,google/gemini-2.5-pro,deepseek/deepseek-v3,qwen/qwen-max. Full list at /dashboard/models.
Continue
VS Code + JetBrains AI assistant
Edit ~/.continue/config.json, point provider:openai at us, list as many models as you like.
Read the guide →Cursor
AI-native VS Code fork
Settings → Models → OpenAI API Key. Set base URL + key, declare custom model names.
Read the guide →OpenWebUI
Self-hosted ChatGPT-like UI
Admin Panel → Settings → Connections. Add Leanroute as an OpenAI-compatible endpoint.
Read the guide →Aider
Terminal-native coding agent
Two env vars, one flag. Drop a .env in your repo and aider routes through us.
Read the guide →Cline
Autonomous VS Code agent
API Provider → OpenAI Compatible. Same base URL + key, pick any flagship model.
Read the guide →n8n
Workflow automation
OpenAI credential + custom base URL. One config covers every Chat Model and AI Agent node in your workflows.
Read the guide →Microsoft Presidio
PII detection + redaction
ML-backed PII anonymization in front of Leanroute. Names, addresses, IDs, account numbers across locales.
Read the guide →Lakera Guard
Prompt injection defense
Hosted SaaS. Detects 'ignore previous instructions' and similar adversarial inputs across model families.
Read the guide →Rebuff
Jailbreak detection
Open source. Multi-layered defense including a distinctive canary-token approach.
Read the guide →Using something we haven't documented? If it accepts a custom OpenAI base URL, it works. Tell us what you're plugging in and we'll add a guide.