Helicone
LLM gateway with logging, caching, and cost analytics
Verdict
Proxy-based approach means zero SDK changes โ add one header and get logging, caching, and rate-limiting. One-line integration is its superpower. Open source under Apache 2.0, with self-hosting and a managed cloud.
Other Observability & Evals
- LangfuseProduction
Open-source LLM engineering platform โ traces, evals, prompts
- Phoenix (Arize)Stable
Open-source AI observability with embedding and RAG tracing
- Weave (W&B)Stable
Weights & Biases' LLM tracing and eval framework
- OpenLLMetryEmerging
OpenTelemetry-based observability for LLMs
- PromptLayerStable
Prompt version control, A/B testing, and analytics platform
- BraintrustEmerging
End-to-end LLM evaluation and experimentation platform