12.8K req/day
760ms latency
0.2% error rate
96.4% accuracy
$820 this month
99.97% uptime (30d)
Usage This Week
14.2% vs prior weekTraining & Evaluation Metrics
Accuracy
96.4%
Precision
94.8%
Recall
92.1%
F1 Score
93.4%
Training dataset2.4M labeled samples
Validation split15% hold-out
Training duration6h 42m on 8×A100
Loss (final)0.0412
Feature Importance
Conversation context length28%
User intent embedding22%
Historical resolution rate18%
Response latency signal14%
Sentiment score11%
Time of day7%
Configuration
Max Tokens
4,096
Temperature
0.7
Rate Limit
500 req/min
Context Window
200K tokens
Top P
0.95
Fallback Model
claude-haiku-4
Deployment History
Version v5.2.1 deployed to production
Jun 14, 2026 · MLOps Team
Passed validation with 96.4% accuracy
Jun 12, 2026 · QA Pipeline
Training run completed · 6h 42m
Jun 09, 2026 · Training Cluster
Dataset frozen for v5.2.1 training run
Jun 02, 2026 · Data Engineering
Model Card
Model ID: claude-sonnet-5
Type: Large Language Model
Provider: Anthropic
Deployed: Jun 14, 2026
Owner: Platform ML Team
Used By
Chatbot8.4K requests/day
Insights Engine4.4K requests/day
Version History
v5.2.1 Current
Jun 14
v5.2.0
May 02
v5.1.4
Mar 18
Cost Breakdown
Input tokens$460.00
Output tokens$310.00
Fine-tuning$50.00
Total$820.00