This is the second time this bug has happened but my original report got buried so I'll make a new one; when attaching an image with no text in the Gemini app and sending just the blank image to Lyria 3 Pro (new chat, model set to thinking) results in a weird output where it attaches what I assume is the internal reasoning/the request that gets sent to the actual model alongside the completed generation. See attached image.
#Lyria 3 Pro text output error.
4 messages · Page 1 of 1 (latest)