CI-NEB shows lower energy for the images than fixed images

[sapper1]

Dear Admin, Proff. Graeme,
I am having very weird problem. My couple of images close to initial and final images show lower energy than the initial images. I tried minimizing those low energy images (CONTCAR) but they go back (very close) to initial and final energy. so, I don't thing those images are in global minima. I followed the CI-NEB steps as you suggested before. initial run with IBRION 3, POTIM 0.01, EDIFFG = -0.1, ISYSM = 0 and then I used IOPT 7 with POTIM 0, EDIFFG -0.03 and IBRION 3 as mentioned in the website. I also put ISYM = 0 for all CINEB runs. CINEB also reaches to required accuracy. I have listed the energy of many repeated runs.
thank you very much
best
Hom

Rxn-cord run-1 run-2 run-3 run-4 run-5
0 -286.74457 -286.74457 -286.74457 -286.74457 -286.74457
1 -287.38759 -287.40687 -287.40725 -287.38982 -287.39033
2 -287.30903 -287.35791 -287.36269 -287.35288 -287.34773
3 -286.80049 -287.01569 -287.04532 -287.13975 -287.04853
4 -286.18302 -286.27159 -286.27678 -286.45376 -286.28527
5 -285.89846 -285.93511 -285.93794 -285.9338 -285.91823
6 -286.51636 -286.83192 -286.80442 -286.6573 -286.77961
7 -287.19867 -287.39988 -287.39418 -287.33355 -287.35608
8 -287.50906 -287.61911 -287.61904 -287.57273 -287.57747
9 -287.61637 -287.64392 -287.644 -287.60012 -287.60079
10 -286.93501 -286.93501 -286.93501 -286.93501 -286.93501
here is my INCAR

Startparameter for this Run:
ENCUT = 400
LCHARG = .TRUE.
LWAVE = .TRUE.
ALGO = Fast
ISPIN = 1
ISTART = 1
LREAL = Auto
IDIPOL = 3
IMAGES = 9
SPRING = -5
Electronic Relaxation 1
EDIFF = 1E-05 :stopping-criterion for ELM
EDIFFG = -0.03
Ionic Relaxation
NSW = 300
IBRION = 3 :ionic relax: 0-MD 1-quasi-New 2-CG
ISIF = 2 :stress and relaxation
POTIM = 0
IOPT = 7
LCLIMB = .TRUE.
ISYM=0
NPAR = 4

[sapper1]

Dear Admin, Proff. Graeme,
here are the initial POSCARS and final CONTCARS and OUTCARS
thank you
best
Hom

[sapper1]

OUTCARS from the last run
thanks

[sapper1]

remaining

[graeme]

I don't see the OUTCARs for the 00 and 10 (reactant and product directories), but the NEB looks just fine. The force on the climbing image (05) is low (<0.01 eV/Ang). Without the endpoint OUTCAR files, I can't address the problem of images having a lower energy, but again, I don't see anything wrong with the NEB.

[sapper1]

Thank you very much Proff. Graeme for the quick reply . here is the OUTCAR-00 and OUTCAR-10
thanks
best
Hom

[graeme]

You have a different k-point mesh for the 00 and 10 images. Use ISYM=0 for the endpoint calculations as well.