diff --git a/.vscode/spellright.dict b/.vscode/spellright.dict
index 62a9a81b4..368799648 100644
--- a/.vscode/spellright.dict
+++ b/.vscode/spellright.dict
@@ -262,4 +262,8 @@ ctx_loader
 ctx_arg_name
 ctx_arg_pos
 dataset
+idx
+subprocess
+usr
+±
 UNPROCESSABLE_ENTITY
diff --git a/ci/benchmark.py b/ci/benchmark.py
new file mode 100755
index 000000000..7cef2e31e
--- /dev/null
+++ b/ci/benchmark.py
@@ -0,0 +1,87 @@
+#!/usr/bin/env python3
+"""
+This file is a script to run bench-marking tests for the FlexMeasures project.
+It can be useful to compare two states, and see if performance has improved or got worse.
+
+Call like this:
+
+./ci/benchmark.py --iterations 5  --sensor 6 --sensor 7 --start 2023-01-01T05:00:00.000Z --duration P1M
+"""
+
+import subprocess
+import time
+import isodate
+import numpy as np
+import tqdm
+import click
+
+# TODO: do not use CLI, but address the underlying functions directly
+
+from flexmeasures.data.schemas import DurationField, AwareDateTimeField
+
+
+@click.command()
+@click.option("--iterations", type=int, default=10)
+@click.option("--sensor", "sensors", type=int, multiple=True, required=True)
+@click.option(
+    "--start",
+    "start",
+    type=AwareDateTimeField(format="iso"),
+    required=True,
+    help="Start queries at this datetime. Follow up with a timezone-aware datetime in ISO 6801 format.",
+)
+@click.option(
+    "--duration",
+    "duration",
+    type=DurationField(),
+    default="P30D",
+    help="Duration of data to be queried, after --start. Follow up with a duration in ISO 6801 format, e.g. PT1H (1 hour) or PT45M (45 minutes).",
+)
+def benchmark(iterations, sensors, start, duration):
+    """
+    This command queries belief data a couple times and tells you how long that took on average (and the STD).
+    """
+    start_str = str(start).replace(" ", "T")
+    start_str = isodate.datetime_isoformat(start)
+    duration_str = isodate.strftime(duration, "P%P")
+    sensor_ids = " ".join([f"--sensor-id {sensor_id}" for sensor_id in sensors])
+    commands = [
+        f"flexmeasures show beliefs {sensor_ids} --start {start_str} --duration {duration_str}",
+        # TODO: add a scheduling operation (wait for result, no queue)
+    ]
+
+    print("I'll be running:")
+    print("\n".join(commands))
+    print()
+    print(f"I'll repeat this {iterations} times.")
+    print()
+
+    timings = {}
+
+    for idx, command in enumerate(commands):
+        for _ in tqdm.tqdm(range(iterations)):
+            start = time.perf_counter()
+            run_command(command)
+            stop = time.perf_counter()
+            try:
+                timings[idx].append(start - stop)
+            except KeyError:
+                timings[idx] = [start - stop]
+
+    # Get mean and standard deviation of timings
+    print(f"All timings are in seconds and iterated {iterations} times")
+    for idx, timing in timings.items():
+        print(f"Command {commands[idx]}: {np.mean(timing)} ± {np.std(timing)}")
+
+
+def run_command(command):
+    # Run command using subprocess and check for errors but don't print output
+    try:
+        subprocess.run(command, shell=True, check=True, stdout=subprocess.DEVNULL)
+    except subprocess.CalledProcessError as process_error:
+        print(process_error)
+        exit(1)
+
+
+if __name__ == "__main__":
+    benchmark()
diff --git a/documentation/changelog.rst b/documentation/changelog.rst
index 9f7ec4054..23361a212 100644
--- a/documentation/changelog.rst
+++ b/documentation/changelog.rst
@@ -19,9 +19,11 @@ New features
 Infrastructure / Support
 ----------------------
 
+- Add benchmark script to test speed of the current code base, running some basic commands. [see `PR #787 <https://github.com/FlexMeasures/flexmeasures/pull/787>`_]
 - Introduce a new one-to-many relation between assets, allowing the definition of an asset's parent (which is also an asset). This hierarchical relationship enables assets to be related in a structured manner. [see `PR #855 <https://github.com/FlexMeasures/flexmeasures/pull/855>`_ and `PR #874 <https://github.com/FlexMeasures/flexmeasures/pull/874>`_]
 - Introduce a new format for the output of ``Scheduler`` to prepare for multiple outputs [see `PR #879 <https://github.com/FlexMeasures/flexmeasures/pull/879>`_].
 
+
 v0.16.1 | October 2, 2023
 ============================