If that is backed up by benchmarks then maybe they should imitate whatever Cursor did. What did they do? They may eventually have to do that. Or they might be

If that is backed up by benchmarks then maybe they should imitate whatever Cursor did. What did they do? They may eventually have to do that. Or they might be starting with an existing Llama model. Maybe I should have said "huge breakthrough or additional dataset".