Side by side
Side by side

Same prefix, different model.

Three pairs that get confused in conversation, laid out row by row. The reading under each table says what the difference amounts to once the specification sheet is out of the way.

ESM-2, ESM3 and ESM C

Three models with the same prefix that answer different questions.
ESM-2ESM3ESM C
What it isMasked sequence encoderMultimodal generative modelMasked sequence encoder
TracksSequenceSequence, structure, functionSequence
GeneratesMasked-token logitsSequence, structure and annotationMasked-token logits
Open sizes8M to 15B1.4B open checkpoint300M, 600M, 6B
Context1024 tokensSee model card2048 tokens
Use it forEmbeddings and baselinesPrompted designEmbeddings at larger scale
Published asScience 2023Science 2025Preprint 2026

ESM3 is the generative line and ESM C the representation line. ESM-2 remains the reference point both are measured against, and it is still the checkpoint most downstream code loads.

AlphaFold 2 and AlphaFold 3

What actually changed between the two.
AlphaFold 2AlphaFold 3
TrunkEvoformer, MSA and pair streamsPairformer, token and pair only
AlignmentCarried through the trunkSummarised early, then set aside
DecoderStructure module, residue framesDiffusion over atoms
OutputOne structure with confidenceSampled structures, then ranked
CoversProteins, and multimers by extensionProteins, nucleic acids, ligands, ions
Triangle updatesYesYes
Published asNature 2021Nature 2024

The triangle machinery survived and the alignment stream did not. Read the difference as a move from one deterministic answer about proteins to a sampled answer about biomolecular complexes.

ESMFold and ESMFold2

The same idea, rebuilt on a newer encoder and a different decoder.
ESMFoldESMFold2
EncoderESM-2, frozenESM C
AlignmentNone at query timeOptional, for difficult targets
DecoderAlphaFold2-style structure moduleAll-atom diffusion
ScopeSingle chainsChains and complexes
Design usePrediction onlyInversion used for binder design
StatusScience 2023Preprint 2026, not peer reviewed

The lineage is a language model feeding a folding decoder. What changed is that the decoder became generative and the alignment came back as an option rather than a requirement.