It really depends on how you configure your OpenVPN/OpenSSH access (which authentication methods are enabled, which type of keys you are using, whether you configured the server properly).
If you enable SSH on your system anyway, using the OpenSSH-based VPN avoids increasing the attack surface (but the performance won't be as great as a datagram-based VPN such as OpenVPN with UDP or wireguard): you'll have more servers to monitor, maintain and upgrade if/when vulnerabilities are found.
On the other hand, if you enable an OpenSSH server only to provide a VPN but fail to disable command execution, an attacker which manages to login will be able to execute commands on your system.
FWIW, I tend to think that configuring and hardening an OpenSSH server is significantly easier than configuring an OpenVPN server. There are quite a few pitfalls in OpenVPN configuration that are easy to make (eg. it is quite easy to configure the clients to accepts other clients as valid server certificates).
When using OpenVPN with TLS 1.2 and below using certificate-based authentication, the certificate client, which often includes the client identity, is sent in clear text. This might be a privacy issue. In contrast, in OpenSSH the SSH user authentication protocol runs on top of the SSH transport protocol: AFAIU, the username is sent encrypted. See the traffic analysis section.