Time how long, on average, it takes to call time.time(), using only time.time() to measure.

time_time.py

#!/usr/bin/env python3
from time import time

# If we iterate a million times,
# then results per-call are in micro seconds.
ITERS = 1_000_000
UNIT = "µs"

# If our first loop below ONLY measures calls to time(),
# then the second loop below (which is to measure loop overhead)
# would be empty, i.e only contain 'pass'. This seems to get
# optimised away, giving incorrect timings, so we call this noop
# from both loops, so that the empty loop is actually executed properly.
def noop():
    pass

# The first loop measures how long to call time(), ITERS times
start = time()
for _ in range(ITERS):
    noop()
    time()
everything = time() - start
print(f"{everything=}{UNIT}")

# The second loop does just the same, but without time(),
# to measure the loop overhead.
start = time()
for _ in range(ITERS):
    noop()
overhead = time() - start
print(f"{overhead=}{UNIT}")

# Output result
result = everything - overhead
print(f"time() calls take {result}{UNIT}")