Pay per routed token + provider costs
Simple pricing
$0.50 per 1M routed tokens. Provider costs passed through at cost. No markup on model pricing.
Global Router
Task-aware routing with automatic fallbacks. Pay only for what you use.
$0.50/ 1M routed tokens
+ provider costs (at cost)
- 35 models (OpenAI, Anthropic, Google, etc.)
- Task classification and model matching
- Automatic retries and fallbacks
- Streaming, function calling, tool use
Enterprise Ready
Custom Brain
Train a custom classifier on your request logs. Better routing for your specific use cases.
$199/ month
Includes 10 Response Memories + 400M Tokens + 10 Training Runs
- Everything in Global Router
- 400M tokens included ($200 value)
- 50% discount on overage ($0.25/1M)
- Upload CSV/JSON request logs
- K-means clustering + transfer learning
- 10 training runs included ($10/run after)
- 100ms routing latency
- Higher Rate Limits
- Attach Custom Response Memories
- Priority support
Technical specs
What you get with ModelPilot routing.
100ms routing latency
Classification and model selection happens in under 50ms. Total latency is routing + model inference.
Data isolation
Custom response memories are isolated per account. Training data is never shared or mixed across users.
OpenAI SDK compatible
Works with any OpenAI client. LangChain, CrewAI, AutoGen, Vercel AI SDK. Change baseURL, keep your code.