You taught the recall. At 6 months old your dog came back across the park. At 10 months old they look straight at you and carry on sniffing. Nothing you built has been erased.
Why it falls apart around 10 months
Around the onset of adolescence, dogs respond less to the cues of the person who raised them, and that responsiveness returns later in the stage. What you are seeing is development rather than a training failure. It is normal, and it is temporary.
Two ordinary things stack on top of that. Your dog can now cover enough ground to get properly gone. And the park pays in smells and movement every second, while you pay once, at the end, when the fun stops.
Three things to stop today
Stop calling when you already know you will not get an answer. An unanswered call spends the cue for nothing. If your dog is 40 feet away and locked onto another dog, walk toward them instead.
Stop recalling only to clip the leash on and go home. Call, feed, then release your dog back to sniffing on 4 of every 5 recalls, so coming back stops predicting the end of the walk.
Put a 20 to 30 foot long line on a harness for the next 2 weeks. It is not a training tool. It is how you stop rehearsing the version where your dog ignores you.
Tonight, indoors, 8 minutes
Recall practice belongs somewhere dull before it belongs at the park. A hallway holds nothing worth sniffing, which makes it the easiest place to make coming back pay well.
Stand 10 feet away in a hallway with 20 small treats in one pocket. Say your recall cue once, in a bright voice, and open your arms.
When your dog arrives, feed 3 treats one at a time, low down beside your knee. Arriving should be worth more than leaving.
Do 6 reps and stop while your dog still wants a seventh. That is the whole session, and it takes about 8 minutes.
Reward-based training is the approach the veterinary behaviour guidance recommends, and reward-based training has been associated with fewer reported problem behaviours. It is also what makes arriving worth repeating.
Then take it outside, in this order
A fenced yard or the dull corner of a field, on the long line, at 10 feet. 6 reps, then stop.
The same spot at 20 feet, once 5 of 6 reps at 10 feet come in quickly and without a detour.
Somewhere busier, back at 10 feet. When you change the place, you drop the distance.
Raise one thing at a time, distance or distraction, never both at once. Staying below the point at which your dog can no longer take food from your hand is the condition for learning outdoors.
Never reel your dog in on the line. Let the line stop them going further, then call once, then back up 2 steps so they chase you in.
Safety note
If your dog's recall changed suddenly, over days rather than weeks, book a veterinary examination before you train anything. Sudden behaviour change warrants a veterinary check, and pain often shows up as a change in behaviour rather than as a limp. The same applies to a dog who has started growling, snapping or freezing.
Why not a remote collar
You will be told a remote collar is faster. A review of aversive training methods found them associated with worse welfare outcomes and no training advantage. Dogs trained with aversive methods showed more stress-related behaviour than dogs trained with rewards. Reward-based methods are what the veterinary behaviour guidance recommends.
How long this takes
No one can promise you a week, and anyone who does is selling something. Responsiveness to your cues comes back later in adolescence, and the practice you do now is what it comes back onto. Expect the hallway to be quick and the park to be slow.
Tonight's version, in one line: 6 reps, in a hallway, 8 minutes, before dinner.