LLM Observability - Requests, Traces and Cost Analytics

Requests

See every request

A complete, searchable ledger of every API call — model, cost, latency, and output — so you can answer any question about what your AI did, and when.

Every call, logged and searchable
Filter, sort, and export in seconds
Reproduce any request in the playground

Traces & Properties

Follow the whole story

Group related calls into traces and tag them with the dimensions you care about — user, feature, environment — to see not just what happened, but why.

Trace multi-step agent flows
Slice by any custom property
Spot patterns across thousands of calls

claude-haiku-4-5-20251001 1229ms $0.0030 200 OK CHAT

Provideranthropic

Input Tokens2,475

Created At8/24/2026, 10:04:55 AM

Output Tokens107

Request ID02355864-7486-4fa8-a3e3-2575b8c9…

Total Tokens2,582

User[email protected]

[email protected] { } JSON

REQUEST

{
▾"model": "claude-haiku-4-5-20251001",
▾"messages": [
▾ {
"role": "system",
"content": "You are a senior customer
support agent for a mobile game company.
\nUsing ONLY the provided context chunks,
draft a helpful, empathetic reply to\nthe
customer's email. If the context is
insufficient, say so honestly rather\nthan
guessing.\nEach context passage is
labelled in brackets with its source type,
version, and\ndate (e.g. [app_change v5.3.0 ·"
}
]
}

RESPONSE

{
"id": "msg_011CeME87W776eNpLQQQ1QPm",
"object": "chat.completion",
"created": 1787558696,
"model": "claude-haiku-4-5-20251001",
▾"choices": [
▾ {
"index": 0,
"message": 
▾ {
"role": "assistant",
"content": null,
"tool_calls": 
▾ [
]
}
}
]
}