Multi-Modal Foundation · biohub

ESMC 300M

Predicts the tokens most likely to fill a masked position.

Verify the licence before commercial useThe mit licence carries terms worth reading before you depend on it.mit

Availability: On demand
Typical latency: 12s
Cost per run: $0.02
Compute tier: S · Micro

Not runnable here yet

We index this model but cannot run it: loads, then fails at inference — see FAILS_AT_INFERENCE.

Everything below is still accurate — the licence, the card and the provenance. Running it needs a loader for this repository, not a bigger machine.

About this model

Predicts the tokens most likely to fill a masked position. Runs on our servers, which pull biohub/ESMC-300M and execute it as a fill-mask pipeline. The input and output shapes were derived from the task the repository declares rather than written by hand — check the model card before relying on a result.

Standardized I/O contract

Every model in the Hub speaks the same contract, which is what lets the Router and the agent call any of them without special-casing.

Inputs

sequencetextarea · required
Sequence with a mask

Outputs

labelstable
Labels with confidence scores, highest first.