LXC supports container metrics within the container, so that the container has independent load avg metrics. This was introduced July 2019 in LXCFS 3.1.2, see https://discuss.linuxcontainers.org/t/lxcfs-3-1-2-has-been-released/5321

On my LXC systems, /proc/loadavg represents expected load for the processes running in each container, whilst the hosts, with many VMs, have an expected higher load as they are handling all systems.

zabbix-agent2, reports the host load on each LXC. I put some examples at https://www.zabbix.com/forum/zabbix-troubleshooting-and-problems/473153-proxmox-8-lxc-container-running-debian-12-zabbix-reports-load-average-of-hostcp -a /

LXC#> uptime; cat /proc/loadavg
 11:56:31 up 11:03,  2 users,  load average: 0.08, 0.04, 0.01
0.08 0.04 0.01 0/681 4114539
ZBX#> zabbix_get -s lxc -k system.cpu.load
3.521484

Note: Tested same issue on zabbix-agentd