Working with secondary-school students in a system where I have a pretty good idea of the quality of OG instruction in the feeder schools, I can usually make a decent determination of whether there is any additional benefit to be obtained from OG by looking at multiple aspects of reading. Fluency is usually the last deficit to resolve, although for some, spelling is where the residual lies. But generally, they can apply their word-level reading skills in isolation, even if slowly.
Keep in mind also that OG addresses primarily the phonological awareness prong of reading decoding. There are two other principal factors: phonological memory and rapid naming (automaticity). When used at a high enough level of intensity and frequency, some of the automaticity may be taken care of through sufficient repetition (though there is another research-based intervention with better data for pure fluency development). Typically, OG-type programs should include at least 2x60 per week, but preferably at least 5x45 in small group instruction. (The home-based individual tutoring programs from All About Learning recommend 5x20 for reading, plus 5x20 for spelling.) The specialists I've worked with have also made decisions about groupings vs individual intervention time based on factors including the student's responsiveness to intervention and level of severity; sometimes even the small group is moving at an incompatible pace for specific students.
Another factor is that there are many more ways to spell a sound than there are ways to pronounce a spelling. It sounds like he actually has made significant progress in OG, as his reading is described as having resolved. It may be that his intervention needs to shift to target spelling/encoding more than reading/decoding at this point. He may know the rules, but he may also need incremental practice to acquire some more fluency.
And one more item: among the more intractable dyslexics, I have seen severe articulation disorders popping up as a pattern. I think it significantly interferes with sound-symbol correspondence, because the sound they generate/hear is not the one aligned with written English. So then it becomes a huge memory task, to match arbitrary sounds with arbitrary symbols. I've seen some success with combining the speech therapy and OG (in one case, the speech path was also our OG therapist, so she was able to actually do both interventions together).
