OverviewLLM calls

LLM call log

Every LLM call across all stages. Click a row to see the prompt, response, retrieved chunks, and retry.

Total calls (1h)
8
Failures
4
50.0%
Total cost (1h)
$0.0959
Total tokens
8,433
⚠
4 calls in the last hour didn't succeed
Inspect each one for the failure type. Common causes: provider rate limits, structured-output parse failures, no approved evidence, timeouts. The runtime automatically falls back to the alternate provider where possible.
A

All calls

Sorted newest-first by default · use filters to find failures.
8 of 8
TimeStageModelOutcomeLatencyTokensCostGrounded?Session
5:34:00 pmdraftinggemini-2.5-prosuccess980ms1212/78$0.01422 citessess_0HXYZK
5:34:00 pmkq_classifiergemini-1.5-flashsuccess142ms84/1$0.0001——
5:18:00 pmdraftinggemini-2.5-prono_evidence1820ms1842/12$0.0218—sess_2P4KQM
5:02:00 pmdraftinggpt-4.1provider_error2840ms1320/0$0.0156—sess_8B1RDT
4:30:00 pmdraftinggemini-2.5-prosuccess1240ms1408/64$0.01683 citessess_3M2VCX
4:30:00 pmjudgegpt-4.1low_confidence1020ms612/48$0.00841 citessess_3M2VCX
4:18:00 pmagent_suggested_replygemini-2.5-prosuccess720ms480/32$0.00581 citessess_3M2VCX
3:30:00 pmdraftinggemini-2.5-protimeout9200ms1240/0$0.0132—sess_5C9XQ1
1–8 of 8
Page 1 of 1
Preview widget