diff --git a/.github/workflows/osrm-backend.yml b/.github/workflows/osrm-backend.yml
index e787b31c6..eabe3b813 100644
--- a/.github/workflows/osrm-backend.yml
+++ b/.github/workflows/osrm-backend.yml
@@ -656,7 +656,7 @@ jobs:
           path: pr
       - name: Install dependencies
         run: | 
-          python3 -m pip install "conan<2.0.0" "requests==2.31.0" "locust==2.28.0"
+          python3 -m pip install "conan<2.0.0" "requests==2.31.0" "numpy==1.26.4" 
           sudo apt-get update -y && sudo apt-get install ccache
       - name: Prepare data
         run: |
diff --git a/scripts/ci/e2e_benchmark.py b/scripts/ci/e2e_benchmark.py
new file mode 100644
index 000000000..4fc900eb9
--- /dev/null
+++ b/scripts/ci/e2e_benchmark.py
@@ -0,0 +1,102 @@
+import requests
+import sys
+import random
+from collections import defaultdict
+import os
+import csv
+import numpy as np
+import time
+import argparse
+
+class BenchmarkRunner:
+    def __init__(self):
+        self.coordinates = []
+        self.tracks = defaultdict(list)
+
+        gps_traces_file_path = os.path.expanduser('~/gps_traces.csv')
+        with open(gps_traces_file_path, 'r') as file:
+            reader = csv.DictReader(file)
+            for row in reader:
+                coord = (float(row['Latitude']), float(row['Longitude']))
+                self.coordinates.append(coord)
+                self.tracks[row['TrackID']].append(coord)
+        self.track_ids = list(self.tracks.keys())
+    
+    def run(self, benchmark_name, host, num_requests, warmup_requests=50):
+        for _ in range(warmup_requests):
+            url = self.make_url(host, benchmark_name)
+            _ = requests.get(url)
+    
+        times = []
+
+        for _ in range(num_requests):
+            url = self.make_url(host, benchmark_name)
+
+            start_time = time.time()
+            response = requests.get(url)
+            end_time = time.time()
+            if response.status_code != 200:
+                if benchmark_name == 'match':
+                    code = response.json()['code']
+                    if code == 'NoSegment' or code == 'NoMatch':
+                        continue
+                raise Exception(f"Error: {response.status_code} {response.text}")
+            times.append((end_time - start_time) * 1000) # convert to ms
+        
+        return times
+    
+    def make_url(self, host, benchmark_name): 
+        if benchmark_name == 'route':
+            start = random.choice(self.coordinates)
+            end = random.choice(self.coordinates)
+            
+            start_coord = f"{start[1]:.6f},{start[0]:.6f}"
+            end_coord = f"{end[1]:.6f},{end[0]:.6f}"
+            return f"{host}/route/v1/driving/{start_coord};{end_coord}?overview=full&steps=true"
+        elif benchmark_name == 'table':
+            num_coords = random.randint(3, 100)
+            selected_coords = random.sample(self.coordinates, num_coords)
+            coords_str = ";".join([f"{coord[1]:.6f},{coord[0]:.6f}" for coord in selected_coords])
+            return f"{host}/table/v1/driving/{coords_str}"
+        elif benchmark_name == 'match':
+            num_coords = random.randint(50, 100)
+            track_id = random.choice(self.track_ids)
+            track_coords = self.tracks[track_id][:num_coords]
+            coords_str = ";".join([f"{coord[1]:.6f},{coord[0]:.6f}" for coord in track_coords])
+            radiues_str = ";".join([f"{random.randint(5, 20)}" for _ in range(len(track_coords))])
+            return f"{host}/match/v1/driving/{coords_str}?steps=true&radiuses={radiues_str}"
+        elif benchmark_name == 'nearest':
+            coord = random.choice(self.coordinates)
+            coord_str = f"{coord[1]:.6f},{coord[0]:.6f}"
+            return f"{host}/nearest/v1/driving/{coord_str}"
+        elif benchmark_name == 'trip':
+            num_coords = random.randint(2, 10)
+            selected_coords = random.sample(self.coordinates, num_coords)
+            coords_str = ";".join([f"{coord[1]:.6f},{coord[0]:.6f}" for coord in selected_coords])
+            return f"{host}/trip/v1/driving/{coords_str}?steps=true"
+        else:
+            raise Exception(f"Unknown benchmark: {benchmark_name}")
+
+def main():
+    parser = argparse.ArgumentParser(description='Run GPS benchmark tests.')
+    parser.add_argument('--host', type=str, required=True, help='Host URL')
+    parser.add_argument('--method', type=str, required=True, choices=['route', 'table', 'match', 'nearest', 'trip'], help='Benchmark method')
+    parser.add_argument('--num_requests', type=int, required=True, help='Number of requests to perform')
+
+    args = parser.parse_args()
+
+    random.seed(42)
+
+    runner = BenchmarkRunner()
+    times = runner.run(args.method, args.host, args.num_requests)
+
+    print(f'Total: {np.sum(times)}ms')
+    print(f"Min time: {np.min(times)}ms")
+    print(f"Mean time: {np.mean(times)}ms")
+    print(f"Median time: {np.median(times)}ms")
+    print(f"95th percentile: {np.percentile(times, 95)}ms")
+    print(f"99th percentile: {np.percentile(times, 99)}ms")
+    print(f"Max time: {np.max(times)}ms")
+
+if __name__ == '__main__':
+    main()
diff --git a/scripts/ci/locustfile.py b/scripts/ci/locustfile.py
deleted file mode 100644
index cd46aaba9..000000000
--- a/scripts/ci/locustfile.py
+++ /dev/null
@@ -1,74 +0,0 @@
-from locust import HttpUser, TaskSet, task, between
-import csv
-import random
-from collections import defaultdict
-import os
-
-class OSRMTasks(TaskSet):
-    def on_start(self):
-        random.seed(42)
-
-        self.coordinates = []
-        self.tracks = defaultdict(list)
-
-        gps_traces_file_path = os.path.expanduser('~/gps_traces.csv')
-        with open(gps_traces_file_path, 'r') as file:
-            reader = csv.DictReader(file)
-            for row in reader:
-                coord = (float(row['Latitude']), float(row['Longitude']))
-                self.coordinates.append(coord)
-                self.tracks[row['TrackID']].append(coord)
-        self.track_ids = list(self.tracks.keys())
-
-    @task
-    def get_route(self):
-        start = random.choice(self.coordinates)
-        end = random.choice(self.coordinates)
-        
-        start_coord = f"{start[1]:.6f},{start[0]:.6f}"
-        end_coord = f"{end[1]:.6f},{end[0]:.6f}"
-        
-        self.client.get(f"/route/v1/driving/{start_coord};{end_coord}?overview=full&steps=true", name="route")
-
-    @task
-    def get_table(self):
-        num_coords = random.randint(3, 100)
-        selected_coords = random.sample(self.coordinates, num_coords)
-        coords_str = ";".join([f"{coord[1]:.6f},{coord[0]:.6f}" for coord in selected_coords])
-        
-        self.client.get(f"/table/v1/driving/{coords_str}", name="table")
-
-    @task
-    def get_match(self):
-        num_coords = random.randint(50, 100)
-        track_id = random.choice(self.track_ids)
-        track_coords = self.tracks[track_id][:num_coords]
-        coords_str = ";".join([f"{coord[1]:.6f},{coord[0]:.6f}" for coord in track_coords])
-        radiues_str = ";".join([f"{random.randint(5, 20)}" for _ in range(len(track_coords))])
-
-        with self.client.get(f"/match/v1/driving/{coords_str}?steps=true&radiuses={radiues_str}", name="match", catch_response=True) as response:
-            if response.status_code == 400:
-                j = response.json()
-                # it is expected that some of requests will fail with such error: map matching fails sometimes
-                if j['code'] == 'NoSegment' or j['code'] == 'NoMatch':
-                    response.success()
-
-    @task
-    def get_nearest(self):
-        coord = random.choice(self.coordinates)
-        coord_str = f"{coord[1]:.6f},{coord[0]:.6f}"
-        
-        self.client.get(f"/nearest/v1/driving/{coord_str}", name="nearest")
-
-    @task
-    def get_trip(self):
-        num_coords = random.randint(2, 10)
-        selected_coords = random.sample(self.coordinates, num_coords)
-        coords_str = ";".join([f"{coord[1]:.6f},{coord[0]:.6f}" for coord in selected_coords])
-        
-        self.client.get(f"/trip/v1/driving/{coords_str}?steps=true", name="trip")
-
-class OSRMUser(HttpUser):
-    tasks = [OSRMTasks]
-    # random wait time between requests to not load server for 100%
-    wait_time = between(0.05, 0.5)
diff --git a/scripts/ci/process_locust_benchmark_results.py b/scripts/ci/process_locust_benchmark_results.py
deleted file mode 100644
index ad71daf7f..000000000
--- a/scripts/ci/process_locust_benchmark_results.py
+++ /dev/null
@@ -1,31 +0,0 @@
-import sys
-import csv
-
-def main(locust_csv_base_name, suffix, output_folder):
-    with open(f"{locust_csv_base_name}_stats.csv", 'r') as file:
-            reader = csv.DictReader(file)
-            for row in reader:
-                name = row['Name']
-                if name == 'Aggregated': continue
-                
-                statistics = f'''
-requests: {row['Request Count']}
-failures: {row['Failure Count']}
-req/s: {float(row['Requests/s']):.3f}req/s
-avg: {float(row['Average Response Time']):.3f}ms
-50%: {row['50%']}ms
-75%: {row['75%']}ms
-95%: {row['95%']}ms
-98%: {row['98%']}ms
-99%: {row['99%']}ms   
-min: {float(row['Min Response Time']):.3f}ms
-max: {float(row['Max Response Time']):.3f}ms
-'''
-                with open(f"{output_folder}/e2e_{name}_{suffix}.bench", 'w') as f:
-                    f.write(statistics)
-
-if __name__ == '__main__':
-    if len(sys.argv) != 4:
-        print(f"Usage: {sys.argv[0]} <locust csv base name> <suffix> <output folder>")
-        sys.exit(1)
-    main(sys.argv[1], sys.argv[2], sys.argv[3])
\ No newline at end of file
diff --git a/scripts/ci/run_benchmarks.sh b/scripts/ci/run_benchmarks.sh
index 7b78ed66c..ffde03bf0 100755
--- a/scripts/ci/run_benchmarks.sh
+++ b/scripts/ci/run_benchmarks.sh
@@ -18,6 +18,7 @@ function run_benchmarks_for_folder {
 
     FOLDER=$1
     RESULTS_FOLDER=$2
+    SCRIPTS_FOLDER=$3
 
     mkdir -p $RESULTS_FOLDER
 
@@ -41,32 +42,27 @@ function run_benchmarks_for_folder {
     measure_peak_ram_and_time "$BINARIES_FOLDER/osrm-customize $FOLDER/data.osrm" "$RESULTS_FOLDER/osrm_customize.bench"
     measure_peak_ram_and_time "$BINARIES_FOLDER/osrm-contract $FOLDER/data.osrm" "$RESULTS_FOLDER/osrm_contract.bench"
 
-    if [ -f "$FOLDER/scripts/ci/locustfile.py" ]; then
-        for ALGORITHM in mld ch; do
-            $BINARIES_FOLDER/osrm-routed --algorithm $ALGORITHM $FOLDER/data.osrm &
-            OSRM_ROUTED_PID=$!
+    for ALGORITHM in ch mld; do
+        $BINARIES_FOLDER/osrm-routed --algorithm $ALGORITHM $FOLDER/data.osrm &
+        OSRM_ROUTED_PID=$!
 
-            # wait for osrm-routed to start
-            curl --retry-delay 3 --retry 10 --retry-all-errors "http://127.0.0.1:5000/route/v1/driving/13.388860,52.517037;13.385983,52.496891?steps=true"
-            locust -f $FOLDER/scripts/ci/locustfile.py \
-                --headless \
-                --processes -1 \
-                --users 10 \
-                --spawn-rate 1 \
-                --host http://localhost:5000 \
-                --run-time 1m \
-                --csv=locust_results_$ALGORITHM \
-                --loglevel ERROR
+        # wait for osrm-routed to start
+        if ! curl --retry-delay 3 --retry 10 --retry-all-errors "http://127.0.0.1:5000/route/v1/driving/13.388860,52.517037;13.385983,52.496891?steps=true"; then
+            echo "osrm-routed failed to start for algorithm $ALGORITHM"
+            kill -9 $OSRM_ROUTED_PID
+            continue
+        fi
 
-            python3 $FOLDER/scripts/ci/process_locust_benchmark_results.py locust_results_$ALGORITHM $ALGORITHM $RESULTS_FOLDER
-
-
-            kill -0 $OSRM_ROUTED_PID
+        for METHOD in route nearest trip table match; do
+            python3 $SCRIPTS_FOLDER/scripts/ci/e2e_benchmark.py --host http://localhost:5000 --method $METHOD --num_requests 1000 > $RESULTS_FOLDER/e2e_${METHOD}_${ALGORITHM}.bench
         done
-    fi
+
+        kill -9 $OSRM_ROUTED_PID
+    done
+
 
 }
 
-run_benchmarks_for_folder $1 "${1}_results"
-run_benchmarks_for_folder $2 "${2}_results"
+run_benchmarks_for_folder $1 "${1}_results" $2
+run_benchmarks_for_folder $2 "${2}_results" $2