Seeded demo for a fictional SaaS — nothing here is a real account. Sign up to see your own

Models

Per-model usage, cost, cache, and responsiveness — measured client-side, as your users feel it.

Model breakdown

View
ModelProviderRequestsErrorsCost% of spendIn tokensOut tokensCache ratep50p95Price breakdownCustom price
gpt-4oopenai84$0.9368.2%202,47141,94021.0%2410ms4007ms
Embedding: 8.1k tok
claude-sonnet-4anthropic281$0.3324.4%49,58212,1979.0%2835ms5322ms
Embedding: 2.0k tok
gpt-4o-miniopenai1624$0.086.0%268,73269,20121.0%895ms1381ms
Embedding: 10.7k tok
claude-3-5-haikuanthropic5$0.021.3%7,9982,8419.0%608ms905ms
Embedding: 320 tok
gpt-4o-internalopenai1$0.000.0%12,40098021.0%1040ms1040ms
Embedding: 496 tok

Provider breakdown

ProviderRequestsShareErrorsCost% of spendIn tokensOut tokensCache ratep50p95
openai24788.2%4$1.0174.2%483,603112,12121.0%1.16s3.74s
anthropic3311.8%1$0.3525.8%57,58015,0389.0%2.66s5.32s
Typical latency (p50)
1.23s
Slow calls (p95)
3.77s

Latency over time

All calls · wall-clock duration

Error rate & reliability

Failures by kind & status

KindStatusCount
invalid_request4002
rate_limited4291
timeout5041
overloaded5291