Skip to content
AnnouncementMajor Advance ↑↑SoftwareMarch 18, 20262 months ago

Figure unveils Helix neural network for cross-task generalization

Figure announced Helix, a generalist vision-language-action neural network enabling Figure 03 to execute cross-task manipulation from visual input without task-specific retraining.

Autonomy Level
Level IILevel III
Human Readiness
Promising Progress
Robovations Score
Unscored
Impact Assessment

Impact on Autonomy L2 to L3 conditional autonomy potential

  • Cross-task manipulation without task-specific retraining
  • Potential L3 conditional autonomy in structured environments
  • Expanded task scope from single-task to multi-task execution
  • Visual reasoning for novel task adaptation

Impact on Readiness Promising Progress unchanged

  • Consumer adoption barrier remains high due to cost
  • Complexity of deployment requires technical expertise
  • Software-only update with no hardware change required
  • Long-term readiness depends on real-world reliability data
Hype vs Reality
Marketing Claim
Documented Reality
End-to-end generalist model enables cross-task generalization
Figure documentation describes Helix as a VLA model trained for multi-task inference on Figure 03. No third-party testing published yet.
Works on real Figure 03 hardware without modification
Manufacturer claims deployment on existing Figure 03 units. Hardware compatibility confirmed in product roadmap; software-only update.
Expands Figure 03 task scope significantly
Claims specific to tasks trained into the model. Task coverage not published. Real-world adaptation limits unknown.
Reduces need for task-specific retraining
Model designed for few-shot adaptation. Evidence of few-shot performance on novel tasks not independently verified.
Bottom Line

Helix represents a declared shift from single-task to multi-task control on Figure 03. Capability claims are significant but require independent performance benchmarking and long-term field data to assess real-world generalization limits.

Technical Details

Model Architecture

Helix is a vision-language-action (VLA) neural network designed for end-to-end control on Figure 03. The model processes visual input and language prompts to generate motor commands without intermediate planning layers. Manufacturer documentation does not specify architecture details, training dataset composition, or model size.

Deployment

Software-only update. No hardware revision required for Figure 03 units. Figure indicates the model runs on onboard compute with latency targets suitable for real-time control. Specific compute requirements and power draw not published.

Training and Generalization

Helix is trained on multi-task data to enable cross-task inference. Manufacturer claims demonstrate execution of manipulation tasks from visual input without task-specific fine-tuning. Performance on out-of-distribution tasks not disclosed. Few-shot adaptation capability mentioned but not benchmarked.

Source Chain

2 sources substantiate this assessment

1Figure AI Official Announcementwww.figure.ai
2025-10-09
2Figure AI Blog
2025-10-09
← Previous entryBoston Dynamics Atlas transitions to fully electric platform for industrial deploymentMar 16, 2026Next entry →Polaris Alpha iQ+ launches with 4-wheel drive smart mappingMar 19, 2026
The Tracker

See all entries in the Tracker.

Open the Tracker

Entry filed Mar 18, 2026 · 2 sources logged · Last reviewed Mar 18, 2026Back to Tracker · Suggest a correction