For Franklin, it probably varied among the languages he knew -- his knowledge of French seems to have been quite good, and there I'd wager that "fluent" might have meant "capable of reading, writing, and having conversations", since he did live in and visit France. It sounds like his spoken French might not have been completely flawless, but all I have to go on there is "Ça ira"-related anecdotes. For Italian, however, he seems to have been able to read that language but not write it, at least to a certain point in this life, but this didn't keep him from corresponding with Italian-speakers. In a letter to Giambatista Beccaria in 1766: "I am pleased to hear that you read English, although you do not write it. I am in the same case with Italian. Hence we can correspond, if this pleases you, more easily if each of us writes his own language." When people describe Franklin as fluent in five languages, they're including both French and Italian, and using fluency in a more general sense of a competent grasp of a language in at least one respect.

Overall, throughout history, what kind of fluency is most useful has probably varied -- some people were at the top of the class as far as modern standards of fluency, but other people who still "knew" that language to some degree may not have been. In most historical contexts I would expect an idea of fluency to put slightly more weight on being able to read and write/compose in a particular language, with the expectation that you'd be more likely to be reading an author in French/Italian/Attic Greek/etc. than living in that language's country of origin and needing to know how to make small talk. There might have been less emphasis on speaking a certain language like a native, or without regional/local accents, but accents and dialects still existed and were remarked upon -- the author of La Vie D'Edouard le Confesseur, an English nun, is pretty up-front about any dialect/grammar trouble she might have as someone to whom French is a second language. She's fluent enough in French to compose lengthy works in it, but still doesn't necessarily write like a highly-proficient native French-speaker. Chaucer's Prioress is described as fluent in French, but it's French "after the scole of Stratford atte Bowe", not French as spoken in Paris. French as spoken in 12th-14th century England had acquired some dialectal quirks of its own and still sounded different to speakers of correct French. Which dialect/accent/set of linguistic conventions is "correct" in any given place and time is a whole other question.

The Prioress instance is just the first one I thought of off the top of my head, and she's unfortunately fictional, but for the discussion of fluency and dialects and the example of the nun of Barking Abbey, check out W. Rothwell's Stratford Atte Bowe and Paris at Jstor. These examples are from a completely different time period/location/social setting than Ben Franklin, and I hope someone can give you some more recent historical answers with a better background than I can. But "knowing a language" has not always meant the ability to write and speak like a native, and I imagine what constitutes ideal fluency has also varied depending on what the individual's main uses for that language were. If they were regularly traveling and having fast-paced conversations with native speakers, for instance, versus if they were reading books in that language, or writing letters. For example: someone who can read and comprehend written Italian very well, compose Italian-language text and comprehend spoken Italian decently, and can't manage to hold an Italian-language conversation to save their life, let alone like a native. This person would not be considered fluent by a long shot in most modern definitions of fluency. But their grasp of Italian is good enough for some purposes, like if they're reading Dante or corresponding with an Italian pen pal and not traveling around Italy. Where those purposes are concerned I can easily imagine another person characterizing them for posterity as "fluent".