@
mint

Real-time metering
for AI usage

See exactly what your AI stack is costing
per workflow and per customer.

JD
usr_6Tni2ma
210% spike
2m ago
Tokens consumed0M
Workflows called0
Total cost$0
Token usage12.2M
research_pipeline
wf_3k9x…m4p2
Completed
5m ago
Trace
web_search
1.2s
summarize
0.8s
extract_findings
1.5s
write_report
2.1s
0s2.8s5.6s
Token usage per task
520 tokens$0.42
340 tokens$0.28
612 tokens$0.61
928 tokens$1.20
Performance
write_report is the bottleneck at 2.1s
write_report uses the most tokens (928)
Cost
write_report costs the most at $1.20
$2.51 total spend this run

One line of code.

No infra changes

Keep it simple. Track what matters.

main.py
from amint import Amint

Amint.init("<API KEY>")

@workflow
def research_pipeline():
    search  = llm.chat(f"Find key facts about {topic}")
    summary = llm.chat(f"Summarize this: {search.content}")
    report  = llm.chat(f"Write a report based on: {summary.content}")

See your first 10 requests

Start tracking in minutes. No card required.

© Hal, Inc.
DocsTermsPrivacy