Where one study disagrees with what other studies report for the same models, and what we are doing to resolve each case.
These are automatic leads, not findings of error. Each release, a check compares how every study ranks the models it shares with other studies. It flags a study that reverses the order the others agree on, or whose gap to a shared reference model lies well outside theirs. Different tasks, populations or metrics often explain a flag.
A flag never hides a result. Flagged results stay in every table, page and download, marked Contested. A result is withheld only when a reviewer marks it disputed.
ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design ranks ESM3 above ESM-2 in 18 of 18 comparisons; pfmbench primary benchmark evidence ranks ESM-2 above ESM3 in 9 of 12 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank ESM-2 above ESM3 in 9 of their 31 comparisons and ESM3 higher in 21; 1 is a tie. Of those papers, ProteinGym ranks ESM3 higher on its own (18 of 19).
Comprehensive assessment of computational algorithms in predicting cancer driver mutations ranks LRT above MetaLR in 15 of 17 comparisons; 2 of 2 other studies rank MetaLR above LRT.
2 papers rank MetaLR above LRT in 12 of their 13 comparisons and LRT higher in 1. Comprehensive assessment… ranks LRT above MetaLR in 15 of its 20 comparisons and MetaLR higher in 2; 3 are ties.
Comprehensive assessment of computational algorithms in predicting cancer driver mutations ranks FATHMM-XF above CADD in 10 of 10 comparisons; 2 of 2 other studies rank CADD above FATHMM-XF.
2 papers rank CADD above FATHMM-XF in all 12 comparisons that are not ties (1 is a tie). Comprehensive assessment… ranks FATHMM-XF above CADD in all 10 of its comparisons.
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks MVP above ESM-1b in 2 of 2 comparisons; 2 of 2 other studies rank ESM-1b above MVP.
2 papers rank ESM-1b above MVP in all 13 of their comparisons. Brandes et al. 2023 ranks MVP above ESM-1b in all 2 comparisons that are not ties (2 are ties).
Zero-shot benchmarking of RNA language models in structural, functional, and evolutionary learning ranks RNA-FM above RNAErnie in 28 of 36 comparisons; 2 of 2 other studies rank RNAErnie above RNA-FM.
2 papers rank RNAErnie above RNA-FM in all 6 comparisons that are not ties (13 are ties). Zero-shot benchmarking of RNA language… ranks RNA-FM above RNAErnie in 28 of its 36 comparisons and RNAErnie higher in 8.
Clinical likelihood ratios and balanced accuracy for 44 in silico tools against multiple large-scale functional assays of cancer susceptibility genes ranks GERP above DANN in 5 of 6 comparisons; 2 of 2 other studies rank DANN above GERP.
2 papers rank DANN above GERP in all 12 comparisons that are not ties (1 is a tie). Clinical likelihood ratios and balanced… ranks GERP above DANN in 5 of its 6 comparisons and DANN higher in 1.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks REVEL above ClinPred in 6 of 7 comparisons; 2 of 2 other studies rank ClinPred above REVEL.
3 papers rank ClinPred above REVEL in 10 of their 12 comparisons and REVEL higher in 1; 1 is a tie. Livesey and Marsh 2025 ranks REVEL above ClinPred in 6 of its 7 comparisons and ClinPred higher in 1.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks ESM-1b above BayesDel in 6 of 7 comparisons; 2 of 2 other studies rank BayesDel above ESM-1b.
2 papers rank BayesDel above ESM-1b in all 10 of their comparisons. Livesey and Marsh 2025 ranks ESM-1b above BayesDel in 6 of its 7 comparisons and BayesDel higher in 1.
Clinical likelihood ratios and balanced accuracy for 44 in silico tools against multiple large-scale functional assays of cancer susceptibility genes ranks MVP above REVEL in 4 of 5 comparisons; 3 of 3 other studies rank REVEL above MVP.
3 papers rank REVEL above MVP in all 19 comparisons that are not ties (4 are ties). Clinical likelihood ratios and balanced… ranks MVP above REVEL in 4 of its 5 comparisons and REVEL higher in 1.
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks ESM-1b above EVE in 6 of 6 comparisons; 2 of 2 other studies rank EVE above ESM-1b.
2 papers rank EVE above ESM-1b in 22 of their 26 comparisons and ESM-1b higher in 1; 3 are ties. Brandes et al. 2023 ranks ESM-1b above EVE in all 6 comparisons that are not ties (2 are ties).
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MutationAssessor above FATHMM-MKL in 7 of 7 comparisons; 2 of 2 other studies rank FATHMM-MKL above MutationAssessor.
2 papers rank FATHMM-MKL above MutationAssessor in 19 of their 25 comparisons and MutationAssessor higher in 6. Livesey and Marsh 2025 ranks MutationAssessor above FATHMM-MKL in all 7 of its comparisons.
Comprehensive assessment of computational algorithms in predicting cancer driver mutations ranks GenoCanyon above SIFT in 6 of 8 comparisons; 3 of 3 other studies rank SIFT above GenoCanyon.
3 papers rank SIFT above GenoCanyon in 14 of their 16 comparisons and GenoCanyon higher in 1; 1 is a tie. Comprehensive assessment… ranks GenoCanyon above SIFT in 6 of its 10 comparisons and SIFT higher in 2; 2 are ties.
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks MetaLR above ESM-1b in 2 of 2 comparisons; 2 of 2 other studies rank ESM-1b above MetaLR.
2 papers rank ESM-1b above MetaLR in all 13 of their comparisons. Brandes et al. 2023 ranks MetaLR above ESM-1b in all 2 comparisons that are not ties (2 are ties).
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks MetaSVM above ESM-1b in 2 of 2 comparisons; 2 of 2 other studies rank ESM-1b above MetaSVM.
2 papers rank ESM-1b above MetaSVM in all 13 of their comparisons. Brandes et al. 2023 ranks MetaSVM above ESM-1b in all 2 comparisons that are not ties (2 are ties).
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks M-CAP above FATHMM-XF in 7 of 7 comparisons; 2 of 2 other studies rank FATHMM-XF above M-CAP.
2 papers rank FATHMM-XF above M-CAP in 22 of their 26 comparisons and M-CAP higher in 3; 1 is a tie. Livesey and Marsh 2025 ranks M-CAP above FATHMM-XF in all 7 of its comparisons.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MetaLR above FATHMM-MKL in 7 of 7 comparisons; 2 of 2 other studies rank FATHMM-MKL above MetaLR.
2 papers rank FATHMM-MKL above MetaLR in 18 of their 25 comparisons and MetaLR higher in 5; 2 are ties. Livesey and Marsh 2025 ranks MetaLR above FATHMM-MKL in all 7 of its comparisons.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MetaLR above FATHMM-XF in 7 of 7 comparisons; 2 of 2 other studies rank FATHMM-XF above MetaLR.
2 papers rank FATHMM-XF above MetaLR in 22 of their 26 comparisons and MetaLR higher in 3; 1 is a tie. Livesey and Marsh 2025 ranks MetaLR above FATHMM-XF in all 7 of its comparisons.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MetaSVM above FATHMM-XF in 7 of 7 comparisons; 2 of 2 other studies rank FATHMM-XF above MetaSVM.
2 papers rank FATHMM-XF above MetaSVM in 22 of their 26 comparisons and MetaSVM higher in 4. Livesey and Marsh 2025 ranks MetaSVM above FATHMM-XF in all 7 of its comparisons.
An openly licensed benchmark and per-gene calibration map for missense pathogenicity predictors on activating cancer drivers ranks BayesDel above REVEL in 6 of 6 comparisons; 2 of 2 other studies rank REVEL above BayesDel.
3 papers rank REVEL above BayesDel in 11 of their 14 comparisons and BayesDel higher in 2; 1 is a tie. Of those papers, ProteinGym ranks BayesDel higher on its own (1 of 1). An openly licensed benchmark… ranks BayesDel above REVEL in all 6 of its comparisons.
Comprehensive assessment of computational algorithms in predicting cancer driver mutations ranks LRT above MetaSVM in 16 of 20 comparisons; 2 of 2 other studies rank MetaSVM above LRT.
2 papers rank MetaSVM above LRT in 12 of their 13 comparisons and LRT higher in 1. Comprehensive assessment… ranks LRT above MetaSVM in 16 of its 20 comparisons and MetaSVM higher in 4.
GENEB leaderboard (leaderboard.json, 13 task-type aggregates) ranks Nucleotide Transformer above Nucleotide Transformer v2 in 37 of 44 comparisons; 2 of 2 other studies rank Nucleotide Transformer v2 above Nucleotide Transformer.
2 papers rank Nucleotide Transformer v2 above Nucleotide Transformer in 33 of their 46 comparisons and Nucleotide Transformer higher in 7; 6 are ties. GENEB leaderboard ranks Nucleotide Transformer above Nucleotide Transformer v2 in 37 of its 45 comparisons and Nucleotide Transformer v2 higher in 7; 1 is a tie.
TraitGym official metrics: AUPRC by chromosome, weighted average (Hugging Face dataset songlab/TraitGym) ranks phyloP above GPN-MSA in 2 of 2 comparisons; 2 of 2 other studies rank GPN-MSA above phyloP.
3 papers rank GPN-MSA above phyloP in 59 of their 67 comparisons and phyloP higher in 2; 6 are ties. TraitGym official metrics ranks phyloP above GPN-MSA in all 2 of its comparisons.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks ESM-1b above MutPred in 6 of 7 comparisons; 2 of 2 other studies rank MutPred above ESM-1b.
3 papers rank MutPred above ESM-1b in 8 of their 11 comparisons and ESM-1b higher in 1; 2 are ties. Of those papers, ProteinGym ranks ESM-1b higher on its own (1 of 1). Livesey and Marsh 2025 ranks ESM-1b above MutPred in 6 of its 7 comparisons and MutPred higher in 1.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MutationAssessor above MPC in 6 of 7 comparisons; 2 of 2 other studies rank MPC above MutationAssessor.
2 papers rank MPC above MutationAssessor in all 16 of their comparisons. Livesey and Marsh 2025 ranks MutationAssessor above MPC in 6 of its 7 comparisons and MPC higher in 1.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks PolyPhen-2 above MVP in 6 of 6 comparisons; 3 of 3 other studies rank MVP above PolyPhen-2.
3 papers rank MVP above PolyPhen-2 in 18 of their 21 comparisons and PolyPhen-2 higher in 2; 1 is a tie. Livesey and Marsh 2025 ranks PolyPhen-2 above MVP in all 6 of its comparisons.
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks GenoCanyon above phyloP in 3 of 3 comparisons; 2 of 2 other studies rank phyloP above GenoCanyon.
2 papers rank phyloP above GenoCanyon in 11 of their 13 comparisons and GenoCanyon higher in 1; 1 is a tie. Brandes et al. 2023 ranks GenoCanyon above phyloP in all 3 comparisons that are not ties (1 is a tie).
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks ESM-1b above VEST4 in 6 of 7 comparisons; 2 of 2 other studies rank VEST4 above ESM-1b.
2 papers rank VEST4 above ESM-1b in 7 of their 10 comparisons and ESM-1b higher in 1; 2 are ties. Livesey and Marsh 2025 ranks ESM-1b above VEST4 in 6 of its 7 comparisons and VEST4 higher in 1.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MutationAssessor above FATHMM-XF in 7 of 7 comparisons; 2 of 2 other studies rank FATHMM-XF above MutationAssessor.
2 papers rank FATHMM-XF above MutationAssessor in 21 of their 26 comparisons and MutationAssessor higher in 5. Livesey and Marsh 2025 ranks MutationAssessor above FATHMM-XF in all 7 of its comparisons.
Comprehensive assessment of computational algorithms in predicting cancer driver mutations ranks GenoCanyon above MetaSVM in 6 of 8 comparisons; 2 of 2 other studies rank MetaSVM above GenoCanyon.
2 papers rank MetaSVM above GenoCanyon in 12 of their 13 comparisons and GenoCanyon higher in 1. Comprehensive assessment… ranks GenoCanyon above MetaSVM in 6 of its 10 comparisons and MetaSVM higher in 2; 2 are ties.
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks SiPhy above phyloP in 3 of 4 comparisons; 2 of 2 other studies rank phyloP above SiPhy.
2 papers rank phyloP above SiPhy in all 13 of their comparisons. Brandes et al. 2023 ranks SiPhy above phyloP in 3 of its 4 comparisons and phyloP higher in 1.
Kedzierska et al. 2025, Additional file 1 (Supplementary Tables S1-S7, xlsx) ranks scGPT above Geneformer in 83 of 89 comparisons; 2 of 2 other studies rank Geneformer above scGPT.
3 papers rank Geneformer above scGPT in 5 of their 7 comparisons and scGPT higher in 2. Of those papers, Parameter-Efficient Fine-Tuning… ranks scGPT higher on its own (1 of 1). Kedzierska et al. 2025 ranks scGPT above Geneformer in 83 of its 97 comparisons and Geneformer higher in 6; 8 are ties.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks SIFT above FATHMM-XF in 6 of 7 comparisons; 2 of 2 other studies rank FATHMM-XF above SIFT.
2 papers rank FATHMM-XF above SIFT in 25 of their 26 comparisons and SIFT higher in 1. Livesey and Marsh 2025 ranks SIFT above FATHMM-XF in 6 of its 7 comparisons and FATHMM-XF higher in 1.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks PolyPhen-2 above FATHMM-XF in 6 of 6 comparisons; 2 of 2 other studies rank FATHMM-XF above PolyPhen-2.
2 papers rank FATHMM-XF above PolyPhen-2 in 20 of their 26 comparisons and PolyPhen-2 higher in 1; 5 are ties. Livesey and Marsh 2025 ranks PolyPhen-2 above FATHMM-XF in all 6 of its comparisons.
Brandes et al. 2023 Supplementary Table 2: full benchmark evaluations of ESM1b and all other methods ranks SiPhy above GERP in 4 of 4 comparisons; 2 of 2 other studies rank GERP above SiPhy.
2 papers rank GERP above SiPhy in 11 of their 13 comparisons and SiPhy higher in 1; 1 is a tie. Brandes et al. 2023 ranks SiPhy above GERP in all 4 of its comparisons.
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MutFormer above DEOGEN2 in 7 of 7 comparisons; An openly licensed benchmark and per-gene calibration map for missense pathogenicity predictors on activating cancer drivers ranks DEOGEN2 above MutFormer in 6 of 6 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank DEOGEN2 above MutFormer in 6 of their 13 comparisons and MutFormer higher in 7. Of those papers, Livesey and Marsh 2025 ranks MutFormer higher on its own (7 of 7).
ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design ranks VESPAl above CARP in 15 of 19 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks CARP above VESPAl in 6 of 6 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank CARP above VESPAl in 10 of their 25 comparisons and VESPAl higher in 15. Of those papers, ProteinGym ranks VESPAl higher on its own (15 of 19).
Clinical likelihood ratios and balanced accuracy for 44 in silico tools against multiple large-scale functional assays of cancer susceptibility genes ranks MetaLR above SuSPect in 4 of 5 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks SuSPect above MetaLR in 6 of 7 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank SuSPect above MetaLR in 7 of their 13 comparisons and MetaLR higher in 5; 1 is a tie. Of those papers, Clinical likelihood ratios and balanced… ranks MetaLR higher on its own (4 of 6).
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks VARITY above MetaRNN in 6 of 7 comparisons; An openly licensed benchmark and per-gene calibration map for missense pathogenicity predictors on activating cancer drivers ranks MetaRNN above VARITY in 3 of 4 comparisons. No other study gives a verdict, so neither is the outlier.
3 papers rank MetaRNN above VARITY in 5 of their 14 comparisons and VARITY higher in 7; 2 are ties. Of those papers, Livesey and Marsh 2025 ranks VARITY higher on its own (6 of 7).
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MutFormer above MPC in 7 of 7 comparisons; An openly licensed benchmark and per-gene calibration map for missense pathogenicity predictors on activating cancer drivers ranks MPC above MutFormer in 6 of 6 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank MPC above MutFormer in 6 of their 13 comparisons and MutFormer higher in 7. Of those papers, Livesey and Marsh 2025 ranks MutFormer higher on its own (7 of 7).
ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design ranks Tranception above ESM-1b in 15 of 19 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks ESM-1b above Tranception in 7 of 7 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank ESM-1b above Tranception in 11 of their 26 comparisons and Tranception higher in 15. Of those papers, ProteinGym ranks Tranception higher on its own (15 of 19).
Clinical likelihood ratios and balanced accuracy for 44 in silico tools against multiple large-scale functional assays of cancer susceptibility genes ranks FATHMM-MKL above SNAP2 in 3 of 4 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks SNAP2 above FATHMM-MKL in 7 of 7 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank SNAP2 above FATHMM-MKL in 8 of their 12 comparisons and FATHMM-MKL higher in 3; 1 is a tie. Of those papers, Clinical likelihood ratios and balanced… ranks FATHMM-MKL higher on its own (3 of 5).
Benchmarking pre-trained genomic language models for RNA sequence-related predictive applications ranks Nucleotide Transformer above RNAErnie in 7 of 7 comparisons; RNAGym fitness leaderboard (v0.1.1), leaderboard/fitness/README.md ranks RNAErnie above Nucleotide Transformer in 4 of 4 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank Nucleotide Transformer above RNAErnie in 7 of their 19 comparisons and RNAErnie higher in 4; 8 are ties. Of those papers, RNAGym fitness leaderboard ranks RNAErnie higher on its own (4 of 4).
GUANinE v1.1 reveals complementarity of supervised and genomic language models ranks DNABERT-2 above PhyloGPN in 6 of 6 comparisons; Genomic heterogeneity inflates the performance of variant pathogenicity predictions (Lu, Liu, Lin and Brandes, bioRxiv 2025, v2) ranks PhyloGPN above DNABERT-2 in 14 of 14 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank DNABERT-2 above PhyloGPN in 6 of their 20 comparisons and PhyloGPN higher in 14. Of those papers, Genomic heterogeneity inflates… ranks PhyloGPN higher on its own (14 of 14).
Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks MutFormer above SIFT in 7 of 7 comparisons; An openly licensed benchmark and per-gene calibration map for missense pathogenicity predictors on activating cancer drivers ranks SIFT above MutFormer in 3 of 4 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank MutFormer above SIFT in 8 of their 13 comparisons and SIFT higher in 3; 2 are ties. Of those papers, An openly licensed benchmark… ranks SIFT higher on its own (3 of 6).
Benchmarking pre-trained genomic language models for RNA sequence-related predictive applications ranks DNABERT above RNA-MSM in 4 of 4 comparisons; Zero-shot benchmarking of RNA language models in structural, functional, and evolutionary learning ranks RNA-MSM above DNABERT in 8 of 8 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank DNABERT above RNA-MSM in 4 of their 20 comparisons and RNA-MSM higher in 8; 8 are ties. Of those papers, Zero-shot benchmarking of RNA language… ranks RNA-MSM higher on its own (8 of 8).
GUANinE v1.1 reveals complementarity of supervised and genomic language models ranks Caduceus above ChromBPNet in 6 of 6 comparisons; Comparative Analysis of Deep Learning Models for Predicting Causative Regulatory Variants ranks ChromBPNet above Caduceus in 4 of 4 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank ChromBPNet above Caduceus in 4 of their 10 comparisons and Caduceus higher in 6. Of those papers, GUANinE v1.1 reveals complementarity… ranks Caduceus higher on its own (6 of 6).
ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design ranks VESPA above CARP in 18 of 19 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks CARP above VESPA in 6 of 6 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank CARP above VESPA in 7 of their 25 comparisons and VESPA higher in 18. Of those papers, ProteinGym ranks VESPA higher on its own (18 of 19).
Clinical likelihood ratios and balanced accuracy for 44 in silico tools against multiple large-scale functional assays of cancer susceptibility genes ranks MetaSVM above Condel in 4 of 5 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks Condel above MetaSVM in 7 of 7 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank Condel above MetaSVM in 8 of their 13 comparisons and MetaSVM higher in 4; 1 is a tie. Of those papers, Clinical likelihood ratios and balanced… ranks MetaSVM higher on its own (4 of 6).
ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design ranks Tranception above ESM-2 in 16 of 18 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks ESM-2 above Tranception in 7 of 7 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank ESM-2 above Tranception in 9 of their 26 comparisons and Tranception higher in 16; 1 is a tie. Of those papers, ProteinGym ranks Tranception higher on its own (16 of 19).
ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design ranks DeepSequence above ESM-1v in 13 of 17 comparisons; Livesey and Marsh 2025, Genome Biology 26:104 (Europe PMC full text XML) ranks ESM-1v above DeepSequence in 6 of 7 comparisons. No other study gives a verdict, so neither is the outlier.
2 papers rank ESM-1v above DeepSequence in 10 of their 26 comparisons and DeepSequence higher in 14; 2 are ties. Of those papers, ProteinGym ranks DeepSequence higher on its own (13 of 19).