Short answer: probably yaknu, but it doesn't matter.
The gemara is comparing kiddushin to the sale of a field, where in the latter case the two parties are a seller and a buyer, and in the former, the groom and the bride's father (in the case of an under-age bride). The gemara claims that the active agent in the sale of a field is the seller, since the verse says ומכר, and he [the seller] sells, whereas the active agent in kiddushin is the groom, since the verse states כי יקח, when he [the groom] takes.
The gemara then brings a refutation from another verse that flips the agent around for the sale of a field, namely Jeremiah 32:44, which has the words יקנו, they will buy. The gemara retorts that we actually read this as יַקנוּ (yaknu), they will sell. This brings the agency back to the seller. If we were to read יִקָּנוּ (yikkanu), they will be bought, the agency would be unclear. Indeed, the Maharsha on this sugya confirms that we chose the correct vowels, and explains: הוא מקנה השדה לקונה, he [the seller] sells [makneh] the field to the buyer.
However, lack of clarity of the agency would be enough to invalidate the gemara's refutation from this verse, and so it seems not to matter how we read it1.
1 The rest of the sugya is also no proof, for if in the gemara's subsequent attempt to re-read יִקַּח (yikkakh), we were to make an analogy to the above and read a hifil form, ie. יַקֵּח (yakkeykh), we would find that we have created a conjugation that is not attested in Tanakh: much better would be to read the hofal form יֻקַּח (yukkakh). Thus, it is difficult to compare the two re-readings to say they have been changed to the same binyan.