Examples

Agent Ops

Search

Search or jump to…

Overview

3 of 5 agents running in production

Tokens today
12.5M+18.2%vs last week
Requests per minute
842+6.4%vs last week
p95 latency
612ms−42 msvs last week
Spend today
$184.32+9.1%vs last week
Token throughput
Thousand tokens per second, by model
Streaming
Model routing
Share of traffic sent to the large model
60%≈ $246/day
QualityCost
Rate limit64% of 10k RPM
  • atlas-large780 ms
  • atlas-fast210 ms
  • atlas-mini95 ms
  • embed-v3degraded
Agents
Click an agent to pause or resume it
  • TriageSupport inboxRunning
    182.4K tok
    atlas-fast · Classifying 42 new tickets
  • ResearcherMarket intelRunning
    964.1K tok
    atlas-large · Summarizing 12 competitor pages
  • CoderPR reviewsRunning
    1.4M tok
    atlas-large · Reviewing #1284 billing refactor
  • ScribeMeeting notesIdle
    88.3K tok
    atlas-fast · Waiting for next recording
  • CollectorInvoice follow-upError
    41.2K tok
    atlas-mini · CRM rate limit, retrying in 30s
Live trace
Tool calls and replies as they happen
  1. 12:00:10 · Triagesearch_tickets(status="new", limit=50)
  2. 12:01:11 · Researcherfetch_url("https://example.com/pricing")
  3. 12:02:12 · CoderThe proration logic misses annual plans. Suggest a test.
  4. 12:03:13 · TriageTagged SUP-3190 as Billing, priority High.
  5. 12:04:14 · Researchervector_search("churn benchmarks 2026", k=8)