back_office_ops · saas · workflow

Grammarly builds a compact on-device spelling and grammar correction model (~1B parameters)

Grammarly's writing assistance goes offline when internet is unavailable because its corrections rely on multiple large models that cannot run on a user's device due to limited memory and processing capabilities.

How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · User writes in real time
Users expect Grammarly to provide real-time suggestions as they write.
Tools used
LlamaT5MLXGQA
Outcome

A compact ~1B parameter model achieves ~210 tokens/second on an M2 Mac with a 70% reduced memory footprint, enabling real-time on-device spelling and grammar corrections without loss in quality.

What failed first

T5, an encoder-decoder model evaluated as a base model candidate, failed the tokenization requirement by converting nonstandard spaces to regular spaces, making it unsuitable.

Results
Time saved50 tokens per second
Volume~210 tokens/second
Source

https://www.grammarly.com/blog/engineering/efficient-on-device-writing-assistance/

How we source this →

Grounding & classification
Source type: technical build writeup
21 fields verified against source quotes.
quality inspectionemailsocial media postfailure mode describedhuman review describedmetric backedproduction runtime claimedtools describedvendor confirmedworkflow describedsoftwarecost reductionthroughput increasetechnical build writeupback office ops