Spend & Reliability Analytics
Real-time correlation between LLM infrastructure costs and operational uptime.
Total Spend
$42,891.20
12% vs last monthReliability Score
99.82%
+0.15% improvementCost per Success
$0.0042
Spike in GPT-4 tokensRetry Rate
1.24%
-0.8% decreaseSpend vs. Reliability over Time
Spend by Model
GPT-4o$24,500.00
Claude 3.5 Sonnet$12,420.00
Gemini 1.5 Pro$4,100.00
Llama 3 (Self-hosted)$1,871.20
Cost vs. Reliability Scatterplot
Analysis of 12 active workflows
Spend by Workflow
| Workflow Name | Primary Model | Success Rate | Latent Fallbacks | Total Cost | Efficiency |
|---|---|---|---|---|---|
| Content Moderation V2 | gpt-4o-mini | 99.98% | 12 | $1,240.45 | Optimal |
| Automated Customer Support | claude-3-5-sonnet | 98.45% | 452 | $15,820.10 | Review |
| Internal RAG Analysis | gpt-4o | 99.10% | 89 | $8,450.00 | Optimal |
| Code Refactoring Engine | claude-3-opus | 96.20% | 1,204 | $12,380.65 | Inefficient |
Reliability-Aware Recommendation
Switching Automated Customer Support from Claude 3.5 Opus to Sonnet with a GPT-4o-mini fallback could save $3,400/mo without degrading performance.