BG95 different registration status via +QCSQ and +QENG

we periodicaly check registration states via +QCSQ, +CEREG, +QENG
sometimes we get conflicting states:
AT+QCSQ
+QCSQ: “NBIoT”,-75,-94,85,-17
OK
AT+CREG?;+CGREG?;+CEREG?
+CREG: 1,1
+CGREG: 2,4
+CEREG: 4,1,“4AAB”,“31FBD5”,9,
OK
AT+QENG=“servingcell”
+QENG: “servingcell”,“SEARCH”
OK

all commands were executed within 200ms
question: what data to trust?

It indicates that the module has successfully registered NBIOT, and you can perform AT+CGPADDR query to see if the IP address can be successfully obtained

Part 2.
Now about param SINR at +QCSQ and +QENG.
Look at documentation about SINR
SINR Integer type. A converted value of SINR. The actual SINR = (1/5) × - 20.
Range: 0–250 (that is, the actual SINR ranges between -20 dB and 30 dB).

AT+QCSQ
+QCSQ: “NBIoT”,-72,-84,129,-11
OK
AT+CREG?;+CGREG?;+CEREG?
+CREG: 1,1
+CGREG: 2,4
+CEREG: 4,1,“4AAB”,“31FBD5”,9,
AT+QENG=“servingcell”
+QENG: “servingcell”,“CONNECT”,“NBIoT”,“FDD”,250,01,31FBD5,130,1711,3,0,0,4AAB,-85,-11,-73,12,
OK

we got
SINR at +QCSQ = 129
SINR at +QENG = 12

does command +QENG response correct data?

we use NIDD/SCEF and dnt have IP context.

Query result to indicate SINR=12

i see, but why +QCSQ responces SINR = 129 differens from +QENG SINR = 12, at same time?

+QCSQ: “NBIoT”,-72,-84,129,-11

means SINR is 5.8 dB according to the algorithm I found:

Values are in 1/5th of a dB. Range: 0–250 which translates to -20 dB to +30 dB

(129/5)-20=5.8

The difference in SINR between the two query results may be caused by fluctuations