Comparison between the results of this work and the SAMPL2 challenge. All values are given in kcal mol⁻¹. The first part of the names in the ‘name’ column refers to the tautomer pair naming convention from the SAMPL2 challenge and in brackets is the name as used in this work. The calculations for Klamt and Diedenhofen¹⁶ were performed with MP2+vib-CT-BP-TZVP, for Ribeiro et al.¹⁵ with M06-2X/MG3S/M06-2X/6-31G(d)/SM8AD, Kast et al.¹⁴ with MP2/aug-cc-pVDZ/EC-RISM/PSE-3, and Soteras et al.³⁵ with MP2/CBS+[CCSD-MP2/6-31+G(d)](d)/IEF-MST/HF/6-31G). On this subset the presented approach was the second best method based on the total RMSE. Bracketed quantities [X,Y] denote 95% confidence intervals.

Name	ΔtGexpsolv [kcal mol−1]	ΔtGcalcsolv [kcal mol−1]	ΔtGcalcsolv [kcal mol−1]	ΔtGcalcsolv [kcal mol−1]	ΔtGcalcsolv [kcal mol−1]	ΔtGcalcsolv [kcal mol−1]
		This work	16	15	14	35
1A_1B (tp_982)	−4.8	−4.7	−4.0	−3.0	−7.7	−4.6
2A_2B	−6.1	−6.8	−5.7	−5.7	−9.7	−6.3
3A_3B (tp_999)	−7.2	−8.4	−7.7	−6.7	−11.2	−7.7
4A_4B	−4.8	−0.4	0.5	0.8	−4.6	0.6
5A_5B (tp_1614)	−4.8	−4.7	−3.9	−4.4	−6.2	−5.6
6A_6B (tp_1005)	−9.3	−11.4	−7.6	−9.7	−11.2	−10.0
RMSE		1.3 [0.7,1.7]	1.4 [0.6,2.1]	1.5 [0.4,2.3]	2.8 [2.0,3.5]	1.3 [0.3,2.1]
7A_7B (tp_141)	6.6	4.9	5.3	6.5	5.1	5.5
10B_10C (tp_1058)	−2.9	−5.3	1.7	0.0	−2.8	2.2
10D_10C (tp_1059)	−1.2	−1.7	3.8	2.6	−0.6	5.0
12D_12C (tp_1514)	−1.8	−2.1	3.3	3.1	−0.8	3.0
14D_14C (tp_1072)	0.3	−1.6	1.9	0.8	0.2	4.0
15A_15B (tp_504)	0.9	6.1	−3.0	3.6	0.0	0.9
15A_15C (tp_505)	−1.2	0.7	−0.8	2.3	−1.9	1.4
15B_15C (tp_506)	−2.2	−5.0	1.8	−1.2	−1.9	0.5
RMSE		2.5 [1.5,3.6]	3.6 [2.5,4.5]	2.9 [1.8,3.8]	0.8 [0.5,1.1]	3.8 [2.5,4.9]
Total RMSE		2.2 [1.4,2.9]	2.9 [2.0,3.6]	2.4 [1.6,3.1]	1.9 [1.2,2.6]	3.0 [1.9,4.0]