You are right that the top right belongs to White with correct play, but according to the programme Sabaki White wins by one point (because they have a captive). It is conceivable, but somewhat unlikely, that White could make enough errors for Black to win.
Sabaki calculates the score like this for area scoring:
and like this with territory scoring:
The top right
To be on the safe side, I plugged your SGF into KataGo (via Sabaki), and it showed me a ko in the top right that Black cannot win: not only do they have hardly any threats, even if they did, they could not finish the ko in their favour.
The sequence that KataGo found was this:
$$Bcm1 Trying for a ko
Unfortunately for Black, White simply has to connect with ➅ at a and Black has no way of finishing the ko to kill White. White can just wait for Black to run out of threats and then capture at c and d.
Black’s only chance is if White plays the ko wrongly, for example:
$$Wcm6 White messes up (9 retakes, 13 @ 7)
- capturing ➅ instead of connecting at ⓫,
- threatening with ➉ instead of the deadlier threat at x,
- missing x again after Black takes at ⓫ and therefore giving up the ko, so
- Black wins by 27 (area) or 26 (territory).