Skip to content

Conversation

ColemanDunn
Copy link

Adds for the o4- to the MODEL_PREFIX_TO_ENCODING dictionary and o4 to MODEL_TO_ENCODING. 4.1 has been added in another PR

Adds support for the `o4-` and `o4` prefixes to the
`MODEL_PREFIX_TO_ENCODING` dictionary.  These prefixes
are now correctly mapped to `o200k_base`.
@taylorn-ai
Copy link

Seems like no one is actually reviewing these PRs... Anyway, I would suggest that you add a test for the o4 models, perhaps similar to what I have done in #407

@hauntsaninja
Copy link
Collaborator

Thanks, this is fixed in recent versions of tiktoken

@kuzminT
Copy link

kuzminT commented Aug 9, 2025

Thanks, this is fixed in recent versions of tiktoken

Thanks for fixing this in recent versions of tiktoken. However, I'm thinking about the approach taken here. There were community PRs addressing these issues that sat without response for months, only to be closed after you implemented your own solution.
While I understand maintainers may prefer their own implementation, completely ignoring community contributions without explanation doesn't align well with open source principles. Contributors invested time and effort into these PRs, and they deserved at least acknowledgment or feedback about why the approach wasn't suitable.
Perhaps it would be helpful to add some guidance in the README about your contribution process - whether you're actively reviewing community PRs or prefer to handle development internally.

@taylorn-ai
Copy link

Couldn’t have said it better myself.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants