A typical API key for a REST-ful application usually happens to be significantly less secure than the access control provided by an OAuth JWT (JSON Web Token) for reasons pertaining to application layer protocol messaging (ordering, syntax, data unit protection--or lack thereof), as opposed to protection resulting only from the use of a particular cryptographic algorithm, mode and/or key size. However, I wouldn't be surprised if API key string generators were more predictable than OAuth access token string generators. API keys are often sent as HTTP GET query parameters within an HTTP request's first line as shown here with the Google Maps JavaScript API:
<script async defer src="https://maps.googleapis.com/maps/api/js key=YOUR_API_KEY&callback=initMap" type="text/javascript"></script>
Due to the fact that the API key string is being passed as an HTTP GET query parameter, it is much easier for intermediate web servers (including proxies), and browsers with client-side scripting languages such as JavaScript or ActionScript to gain read and/or write access to the API key. Compare this to other types of HTTP actions such as PUT or POST where the query parameters are more tightly concealed from the aforementioned technologies. Almost all web server software will write the src attribute value in the script tag above to access_log and/or error_log files including the API key, as the query parameter variable values are part of the CGI (Common Gateway Interface) environment variables: SCRIPT_PATH and QUERY_STRING. Refer to CWE-598 for more information.
OAuth access tokens on the other hand, are generated on a per-session basis. Being granted an access token by a secure authentication provider will not occur, until the provider has received proof that the requesting user is entitled to requested privileges; such proof might be established through knowledge of credentials (i.e. a corresponding username and password pair.) Other times, access control might be more restrictive and access tokens are only provided for a small sub-set of privileges within a particular app/site/API sub-component, area of operation, control sphere, etc. The permissions ultimately granted to the end user can be as fine-grained as the system administrator wishes. Note that access tokens are programmed to expire after a set amount of time and are capable of providing discretionary access control between various users/groups, privileges/capabilities, etc.
Access tokens are often transferred outside of the URL in the HTTP request header's Authorization field, for example. Sometimes, custom authentication framework implementations will cause the token to be transmitted within a cookie that has the HttpOnly, Secure and SameSite flags enabled--or as a custom HTTP request header such as X-Auth-Token as publicly documented for Oracle's Cloud Storage SaaS: Oracle's Cloud Storage Service API:
It is extremely rare for HTTP request headers and cookie values to be logged by web browser/server software; they're also more difficult to access programatically due to CORS (Cross Origin Resource Sharing.) In comparison, the API keys passed as HTTP GET parameters can be extracted with client-side JavaScript from the DOM (Document Object Model).
For these reasons, the complexity required to obtain access tokens from an authentication framework such as OAuth is much higher than what is needed to log the usage of an API key. Furthermore, the robustness of authentication and authorization frameworks allows the access token to be encapsulated within the HTTP protocol in ways that it is rather difficult to view or tamper with the token.