Weflayr Documentation
LLM observability for OpenAI and Mistral AI. Drop-in SDK wrappers that stream cost, latency, and error events to your dashboard with zero impact on your existing code.
Python 3.13+
OpenAI
Mistral AI
What is Weflayr?
Weflayr instruments your AI API calls and streams telemetry events to a central intake API. Drop-in SDK wrappers that stream cost, latency, and error events to your dashboard with zero impact on your existing code.
How it works
Every instrumented call fires three events:
| Event | When | Contains |
|---|---|---|
<call>.before |
Before the LLM call | model, message count, tags |
<call>.after |
On success | elapsed_ms, token usage, tags |
<call>.error |
On failure | elapsed_ms, error type, status code |
Events are posted fire-and-forget in a background thread — they never block your main execution path.
Supported providers
| Provider | Import | Clients |
|---|---|---|
| OpenAI | weflayr.sdk.openai.client |
OpenAI, AsyncOpenAI |
| Mistral AI | weflayr.sdk.mistralai.client |
Mistral |