He can say it faster, but however fast you go it never sounds exactly like the word, so I understand your brain has to make a little leap to fill that gap. When you say, "what's the word?", he will say something with those sounds in like 'matter' or make up a word "malitobobby'. My favourite was 'o-n' - he said it was 'onion'!!! But after reading it a few times, he knows it's 'on' now, so he doesn't have to blend it anymore. Which is why I was wondering whether teaching him visual recognition will help him. However, if he is given a word like 'blink', he won't be able to sound it and then blend it.
I often wonder why people think that learning letter/sound correspondences isn't 'visual recognition'. Why not? They are not smelling the letters, or feeling them; they are seeing them. There is no physical difference between seeing a letter or group of letters and saying the sound they represent and seeing a group of letters and saying the word that they represent.
There is a difference in cognitive loading, though, as memorising whole words requires far more 'memory' in remembering the shape and letter sequence in the word and in the sheer number of words to be memorised. There are some 250,000 words in a standard English dictionary and it is absolutely impossible for any individual (apart from the very exceptional types who can memorise an entire phone book!) to memorise each individual word. Whereas in learning phonics the load is reduced to learning about 160 -180 common letter/sound correspondences.
I would very strongly recommend that you persist with the decoding and blending and don't try teaching words as 'wholes'. It is not uncommon for children to have a problem with decoding but with persistence and patience the penny usually drops - I work with secondary age children and not one of them is unable to decode and blend, so they all must have 'got it' at some time.
Practical (thought you may know all this already)
Make sure he is using 'pure' sounds (which he should be if the Miskin programme has been properly taught), putting an /uh/ sound on the end, as in /cuh/ /a/ /tuh/ distorts the sounds and makes the word very difficulty to recognise.
Instead of trying to say the sounds fast, try saying them slowly and letting one 'slide' into the next. I know that the Miskin programme teaches 'robot talk' which doesn't suit all children who are learning to blend.
Make sure that he is still 'reading' the sounds as he blends. Many children decode the sounds, then look away to try blending (trying to memorise them). When they are not looking at what is on the page they can tend to put in sounds which aren't there and produce odd words.
Can he blend a word which you say in 'sounds'. If you said 'Where is your /c/ /oa/ /t/ would he be able to recognise the word and tell you the answer?
Can he orally break a word into its component sounds?
Has he had his hearing and sight checked?
The BRI books are well worth trying as they are 'books' but they take the process very slowly and allow for lots of practice.