Model comparison · robotics

SmolVLA 450M vs pi 0.7

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
Physical Intelligence · activepi 0.7Verified Aug 28, 2026

Technical differences

Side-by-side facts

Indexable
FieldSmolVLA 450Mpi 0.7
DeveloperHugging Face LeRobotPhysical Intelligence
FamilySmolVLApi
ModelSmolVLA 450Mpi 0.7
Version450M0.7
Lifecycleactiveactive
Input modalitiesText, Image, Robot stateText, Image, Robot state
Output modalitiesRobot actionRobot action
Context windowUnknownUnknown
Total parameters450000000Unknown
Active parametersUnknownUnknown
Licenseapache-2.0Unknown
Open weightsYesNo
API availableNoUnknown
Self-hostableYesUnknown
Capabilitiesasynchronous-inference, fine-tuning, low-cost-hardware, manipulationcross-embodiment, dexterous-manipulation, language-steering, visual-subgoals
Robotics model typeVision-language-action modelVision-language-action model
Action representationContinuous action chunks from a flow-matching action expertContinuous robot actions conditioned by multimodal prompts
Control architectureSmolVLM2 backbone with flow-matching action expertHigh-level policy, world model, and action expert
Inference locationOn deviceUnknown
Native control rate (Hz)UnknownUnknown
Supported embodimentsSO-100, SO-101, LeKiwi, LIBERO Frankamobile manipulators, bimanual UR5e, multiple fixed manipulators
Training dataCompatibly licensed LeRobot community datasets totaling fewer than 30,000 episodes in the cited release.Robot demonstrations, autonomous data, egocentric human data, and multimodal web data described by the publisher.

15 comparable fields · 11 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

pi 0.7 capabilities

cross-embodimentdexterous-manipulationlanguage-steeringvisual-subgoals
Input price
Output price
Serving providers0
Canonical IDphysical-intelligence/pi-0.7

Internal comparison graph

Related comparisons

All robotics comparisons →

Primary evidence

Sources and freshness