Maximum Model Size | Vantage BYOM - Maximum Model Size - Teradata Vantage

Teradata Vantage™ - Bring Your Own Model User Guide

Deployment
VantageCloud
VantageCore
Edition
VMware
Enterprise
IntelliFlex
Lake
Product
Teradata Vantage
Release Number
7.0
Published
October 2025
ft:locale
en-US
ft:lastEdition
2025-11-07
dita:mapPath
fee1607120608274.ditamap
dita:ditavalPath
ayr1485454803741.ditaval
dita:id
fee1607120608274

The maximum model size is 2GB, the maximum size of a BLOB value. Consider there is an expansion factor that varies per model when loaded into memory. See Memory Cache Size for more information. For better performance, Teradata recommends model sizes lower than 200MB.

H2O

H2O models exceeding 100 MB are verified to make sure there is enough native memory available on the system to run the model properly. H2O models expand when read into memory by a factor of up to 2.75 times. Verifying the model includes taking the amount of NON FSG Cache native memory and making sure it exceeds the input models size x 2.75. If the system does not exceed the input model size times 2.75, an error is displayed.

For example, if your node has 500 MB of NON FSG cache memory, no error occurs with a model size of 181 MB or less.
input model size x 2.75 =  < node size
181 x 2.75 = 497.75 < 500

You can disable the memory verification by setting the enableMemoryCheck argument to false: EnableMemoryCheck ('false'). Note that if there is not enough native memory available for the model, system performance will be impacted.

ONNX

ONNX models exceeding 33MB are verified to make sure there is enough native memory available on the system to run the model properly. ONNX models expand when read into memory by a factor of up to 2.75 times. Verifying the model includes taking the amount of NON-FSG cache native memory and making sure it exceeds the input models size x 2.75. If the system memory does not exceed the input model size times 2.75, an error is displayed.

For example, if your node has 500MB of NON-FSG cache memory, no error occurs with a model size of 181MB or less.
input model size x 2.75 =  < node size
181 x 2.75 = 497.75 < 500

You can disable the memory verification by setting the enableMemoryCheck argument to false: EnableMemoryCheck ('false'). Note that if there is not enough native memory available for the model, system performance will be impacted.

DataRobot

The maximum model size for DataRobot is 2GB (the uncompressed size of the model as reported by DataRobot AI). You can consult this information the DataRobot AI section Models > Leaderboard > name of your model > Describe > Model Info > Model File Size. Teradata recommends model sizes lower than 200MB.