Helix 02 vs Hy4 preview

At a Glance

Compare
Helix 02Figure
Pricing and Limits
Context windowMaximum documented tokensNot reported1,000K
Model facts checkedAug 29, 2026View model evidence →Sep 2, 2026View model evidence →
Different Model RolesThese models do not share a sourced market category. Their primary-source facts remain comparable below, while performance claims require matched evidence.

Available Benchmarks

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.

Side-by-Side Facts

FieldHelix 02Hy4 preview
DeveloperFigureTencent
FamilyHelixHy4
ModelHelix 02Hy4 preview
Version02Hy4 preview
Lifecycleactivepreview
Released2026-01-052026-08-28
Knowledge cutoffUnknownUnknown
Input modalitiesText, Image, Robot stateText
Output modalitiesRobot actionText
Context windowUnknown1,000K
Total parametersUnknown770B
Active parametersUnknown49B
LicenseUnknownapache-2.0
Open weightsNoYes
API availableNoYes
Self-hostableNoYes
Provider accessUnknownOpenrouter (Standard)
Capabilitiesdexterous-manipulation, long-horizon-control, tactile-control, whole-body-controlchat, generation, reasoning, tools
Robotics model typeVision-language-action modelUnknown
Action representationFull-body joint targetsUnknown
Control architectureSemantic reasoning, visuomotor policy, and kHz whole-body controllerUnknown
Inference locationOn deviceUnknown
Native control rate (Hz)UnknownUnknown
Supported embodimentsFigure 03Unknown
Training dataFigure reports more than 1,000 hours of human motion data plus sim-to-real reinforcement learning for its whole-body controller.Unknown

Helix 02 Capabilities

dexterous-manipulationlong-horizon-controltactile-controlwhole-body-control
Model typeVision-language-action model
InferenceOn device
Action representationFull-body joint targets
Supported embodiments1
Canonical IDfigure/helix-02

Hy4 preview Capabilities

chatgenerationreasoningtools
Model typeUnknown
InferenceUnknown
Action representationUnknown
Supported embodimentsUnknown
Canonical IDtencent/Hy4-preview

Primary Evidence

Sources and Freshness

Questions

Helix 02 vs Hy4 preview FAQs

Is Helix 02 or Hy4 preview better for coding?+

This comparison does not currently contain a protocol-matched coding benchmark for both Helix 02 and Hy4 preview, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.

Which is cheaper, Helix 02 or Hy4 preview?+

Only Hy4 preview has a directly sourced input price: $0.834 per million tokens. Only Hy4 preview has a directly sourced output price: $2.501 per million tokens.

Which has a larger context window, Helix 02 or Hy4 preview?+

Neither model has a larger sourced context window in this comparison. Helix 02 is — and Hy4 preview is 1,000K.

Which performs better in benchmarks, Helix 02 or Hy4 preview?+

There is no overall benchmark winner: At least two independently verified, protocol-matched benchmarks are required for an overall winner.

Can Helix 02 or Hy4 preview be self-hosted?+

Hy4 preview is the only model in this pair currently marked as self-hostable. Helix 02 is not marked open weight; Hy4 preview is open weight.

Can Helix 02 and Hy4 preview understand images?+

Helix 02 is documented with image input; Hy4 preview is not documented with image input. This reflects supported input modalities, not vision quality.

Which can generate longer answers, Helix 02 or Hy4 preview?+

Neither has a larger sourced maximum output. Helix 02 is — and Hy4 preview is —.

Do Helix 02 and Hy4 preview support reasoning and tool use?+

Helix 02: image input. Hy4 preview: reasoning and tool calling. Feature support does not establish relative quality.

Which is available from more inference providers, Helix 02 or Hy4 preview?+

Helix 02 has 0 sourced provider routes; Hy4 preview has 1, so Hy4 preview has broader tracked availability.

Which offers better value, Helix 02 or Hy4 preview?+

There is no universal value winner. Compare the input and output prices above with the matched benchmark result for your workload: cheaper tokens can be offset by different quality, token usage, latency, or provider availability.

Send Feedback