Model comparison · robotics

SmolVLA 450M vs GR00T N1.7 3B

Symmetrical, primary-source facts with unknown values left visible. No winner is assigned without comparable measurements.

FieldAt a glance
Hugging Face LeRobot · activeSmolVLA 450MVerified Aug 28, 2026
NVIDIA · activeGR00T N1.7 3BVerified Aug 28, 2026

Technical differences

Side-by-side facts

Indexable
FieldSmolVLA 450MGR00T N1.7 3B
DeveloperHugging Face LeRobotNVIDIA
FamilySmolVLAIsaac GR00T
ModelSmolVLA 450MGR00T N1.7 3B
Version450MN1.7
Lifecycleactiveactive
Input modalitiesText, Image, Robot stateText, Image, Robot state
Output modalitiesRobot actionRobot action
Context windowUnknownUnknown
Total parameters4500000003000000000
Active parametersUnknownUnknown
Licenseapache-2.0apache-2.0
Open weightsYesYes
API availableNoNo
Self-hostableYesYes
Capabilitiesasynchronous-inference, fine-tuning, low-cost-hardware, manipulationcross-embodiment, dexterous-manipulation, whole-body-control, fine-tuning
Robotics model typeVision-language-action modelVision-language-action model
Action representationContinuous action chunks from a flow-matching action expertPredictive chunks of relative joint motions
Control architectureSmolVLM2 backbone with flow-matching action expertVision-language backbone with action expert
Inference locationOn deviceOn device
Native control rate (Hz)UnknownUnknown
Supported embodimentsSO-100, SO-101, LeKiwi, LIBERO FrankaUnitree G1, AgiBot Genie-1, Fourier GR-1, bimanual manipulation platforms
Training dataCompatibly licensed LeRobot community datasets totaling fewer than 30,000 episodes in the cited release.Mixture of real teleoperation, synthetic robot data, and internet-scale video described by NVIDIA.

19 comparable fields · 10 material differences · Pair passes the primary-source comparison gate

Primary benchmark evidence

Comparable performance

All benchmark results →
No protocol-matched benchmark yet.Results appear here only when both models share the same benchmark version, metric, evaluation protocol, and evidence class.

SmolVLA 450M capabilities

asynchronous-inferencefine-tuninglow-cost-hardwaremanipulation
Input price
Output price
Serving providers0
Canonical IDlerobot/smolvla_base

GR00T N1.7 3B capabilities

cross-embodimentdexterous-manipulationwhole-body-controlfine-tuning
Input price
Output price
Serving providers0
Canonical IDnvidia/GR00T-N1.7-3B

Internal comparison graph

Related comparisons

All robotics comparisons →

Primary evidence

Sources and freshness