Loading

Preparing your page

AI Stats is fetching the latest data for this page. This usually only takes a moment.

Return to the main page and start from a clean slate.

Explore the catalogue of AI models and their details.

Check the documentation for guides, references, and examples.

Explore the Gateway

Access the largest unified AI Gateway and use over 300+ models easily.

If this screen doesn't disappear after a short while, you can refresh the page or use one of the links above to continue.

Claude 3 Haiku Performance - Latency & Token Trajectory | AI Stats

Throughput

32.38 t/s+113.2%

Latency

571 ms-15.9%

E2E latency

772 ms-11.5%

Provider performance

Last 24 hours

				Uptime
Anthropic	32.38 t/s	571 ms	772 ms	No data during Requests from Tue 00:00 to Tue 06:00 100% success during Requests from Tue 06:00 to Tue 12:00 No data during Requests from Tue 12:00 to Tue 18:00 No data during Requests from Tue 18:00 to Wed 00:00 No data during Requests from Wed 00:00 to Wed 06:00 100.0%

Model Uptime

Time-of-day performance

Throughput by hour

All times in UTC. Aggregated from hourly medians captured during the selected reporting window.

Token usage growth

Cumulative tokens since launch

Release date: Mar 13, 2024

Deprecation

Lifecycle status

This model has not been marked deprecated.

Token milestones

Days to hit key thresholds

1.0M tokens	—	—
10.0M tokens	—	—
100.0M tokens	—	—
1.0B tokens	—	—

Successor models

Release milestones

Claude 3.5 Haiku

236d

Nov 4, 2024