Did 19.18 resolve those HR inaccuracies introduced with SW Version 18? Is it safe to update from 17.26?

I have used all subsequent versions (Release, non Beta) and even since 17.26 and have had no problems. I understand that several people have reported problems, I don't know if they are due to hardware versions or what, but I also believe that they are isolated cases in the ratio of number of devices vs. reports made. Now I'm on 19.18

7457844 said:

Would you say that the current version fixed what went wrong since?

No.

17.26 was impacted from pulse incontinent too. 

I was briefly on 18.22 and 18.23. With these version my HR for walking at steady fast pace as reported by the watch was around my resting HR.

I went back to 17.26 and the measured HR is normal again. I will stay on this version. I don't understand the latest developments around OHR: always on, warming up before run/walk/hike activities, dynamic switching OHR/HRM. All these popped up about 2 years after the 955 was released. The watch was fine for me until then. Other users on this forum may have different views.

With 17.26 there are UI bugs but nothing major for me. As for the new "features" I don't really care about: Sleep Coach, JetLag Advisor, etc.

My advice: stay on 17.26. I know I will.

TL;DR: Initial impression, seems to be all OK.

First a bit of personal Optical Heart Rate history to judge the value of my answer and experience:

7th of July 2017. Maiden run with a Fenix 5X which contained an OHR. Immediate cadence lock, but after tightening the watch strap further to stabilize the heavy piece, my pulse there agreed with a Forerunner 630 (lacking OHR) connected with a Garmin HRM-Run chest strap. Further test runs confirmed that the OHR consistently, and for all kinds of runs, stayed 1-2 bpm higher than the strap and that the delay in displayed pulse changes, in relation to physical perception, was ca 5-6 seconds (strap delay 2-3 seconds). Cold winter weather could cause all kinds of errors - dips or spikes - so it bacame a habbit to use only chest straps from October to May.

29th of January 2022. First run with a Coros Pace 2 (OHR) and the Fenix 5X now connected to a Polar H10 chest strap. The Coros watch stayed 1-2 bpm above the strap and had the same 5-6 seconds delay as the Garmin OHR. There was one spike episode during that slightly cold (+6C) run, but warmer weather runs showed the Coros OHR to be as reliable as the Garmin version.

7th of September 2023. First run with a FR955 on my left arm, the Pace 2 on right arm - no chest strap. 10.65 km on both watches, average 144 bpm and max 167 bpm with Garmin, 144 and 167 with Coros... Slight current pulse variations between the watches during the run, as should be expected. The FR955 was on Firmware 16.17 at the time, fully updated from whatever came at sale. I raced in a Half Marathon the 30th of that month using the OHR of both watches: Garmin avg 154 max 171, Coros 154 and 168.

I updated the FR955 to FW 17.26 on the 10th of January 2024 but used only the Polar H10 strap due to temperature all the way until the 30th of April when we finally began the 'nice' season here (that day +16.3C -- +18.0C during the run). Since I always run with both the FR955 and the Coros Pace 2 I could tell that there was nothing _obviously_ erroneous with the new firmware in regards to OHR accuracy. At least for me, who am skinny with very little melanin to block the sensor. But of course I've stayed clear of the 18.xx series of firmwares due to forum reports of poor OHR function and other bugs of importance to my usage.

A few days ago, the 31 of May, I had updated the FR955 to FW 19.18 and went on the following run from a book: "Fartlek Run: 11 km Easy w/8x40s Intervals @10K-3K-Pace". Scrutinizing the pulse curve afterwards, both in the Garmin and the Coros app, it followed were closely both the effort and the terrain undulations. Garmin avg 148 bpm max 166 (It was, for me, very hot 18.9C -- 20.9C), Coros 147 and 167.

Based on this run, I don't expect any major OHR problems henceforth with this firmware. Other known bugs like DSW etc are of no concern to me since the watch sleeps in a closet between exercises...

The 19.18 release is definitely buggy. Spotify does not work at all currently so I would hold off on updating. I have heard nothing from Garmin around this issue and it's really frustrating not having a core feature like this

First of all - thanks for the detailed report. :) Back to built in OHRS - i speak about my experience only. 

A very first OHRS failure happens to me @ early 17.xx FW - it was an easy run - after some high load days. My HR jumps from normal 100 to crazy 140 - i was close to canceling my session, but wait... I checked my HR manually and it was 90-100 as expected. After the next 10 min a pulse jumped back to normal 100-110 and it was all OK again. 

The next serious case was 2 months later, in May - on a mountain walk - i got a HR higher than my HRmax(about 180) - 195-205 for  a 10-15 min continuously - i'v changed a hand, took it off - nothing helps - it was really weird. I thought it was doubling my HR. It was over 20C and after 2 hours of moving... The biggest mistake that you can make in such a case - and I've done it - save a crappy exercise after all - My TE was red and I got overloaded.. Very uncomfortable.  

What would I like to say by this? This happens quite rarely, but this is definitely a new error, before, and even on older versions, there were deviations associated with the features of this type of sensor - temperature influence, delayed readings, cutting off the maximum on the sprint - but what is happening now makes the use of this watch unreliable. I'm increasingly wondering - maybe I should do sports without a watch at all? 

Good afternoon How did you go back to the older firmware? Please describe the process

Maybe this post and its thread will help you:
forums.garmin.com/.../1724330