Available integrations
Providers
- OpenAI - GPT-4 and other OpenAI models
Gateways
- OpenRouter - Universal LLM gateway with load balancing
- LiteLLM - Unified interface for 100+ LLMs
- Portkey - AI gateway with routing and fallbacks
- Helicone Gateway - AI gateway with routing and fallbacks
Observability
- Helicone - LLM observability and monitoring
- Langfuse - LLM engineering platform
- LangSmith - LangChain debugging and monitoring
- Weights and Biases - ML experiment tracking
Narev is the missing experimentation layer
A common LLM stack has three layers:| Layer | Tools | Purpose | |
|---|---|---|---|
| OpenAI, Claude, Groq | Generates responses | Helps find optimal configuration | |
| OpenRouter, LiteLLM, Portkey | Provides common interface | ||
| LLM: Helicone, Langfuse, W&B | Shows LLM history | ||
| FinOps: Vantage, CloudZero, Finout | Tracks costs and spend |