{"id":"9fc8ad4b-3d2f-4a3e-a0fe-f7ea7e138eb2","slug":"helix","name":"Helix (and Helix-02)","description":"Figure AI's onboard dual-system VLA. System 2 is an internet-pretrained VLM (7B params, 7-9 Hz) handling scene and language understanding; System 1 is a fast reactive visuomotor policy (80M params, 200 Hz). Helix-02 (January 27, 2026) extended to full-body control via System 0, a 1 kHz neural prior trained on 1,000+ hours of human motion data that replaced roughly 109,504 lines of hand-engineered C++. The first VLA to control a full humanoid upper body including individual fingers from one set of weights, and the first to run two robots from one set of weights.","brainType":"foundation-model","isOpen":false,"maturityStage":"commercial","architecture":"Dual-system VLA. System 2 (S2): onboard internet-pretrained VLM, 7B params, 7-9 Hz. System 1 (S1): fast reactive visuomotor policy, 80M params, 200 Hz. Helix-02 added System 0, a 1 kHz neural prior trained on 1,000+ hours of human motion data, replacing roughly 109,504 lines of hand-engineered C++. Runs on embedded low-power onboard GPUs, no cloud.","reviewStatus":"reviewed","sources":[{"url":"https://www.figure.ai/news/helix","label":"Figure AI: Helix (primary)"},{"url":"https://www.figure.ai/news/helix-logistics","label":"Figure AI: Helix in logistics"},{"url":"https://www.techtimes.com/","label":"TechTimes: Helix-02 8-hour autonomous shift (May 14, 2026)"},{"url":"https://luigifreda.com/","label":"Luigi Freda: Helix architecture overview"},{"url":"https://techstartups.com/2025/02/21/ai-startup-figure-unveils-helix-new-ai-model-lets-robots-follow-voice-commands-handle-unknown-objects/","title":"Figure unveils Helix VLA; signals confidence in in-house AI after the OpenAI split (Feb 21, 2025)","sourceName":"Tech Startups"},{"url":"https://newatlas.com/robotics/helix-vla-figure-02-robot/","title":"Figure's Helix VLA: humanoids do tasks they were not trained for (Feb 2025)","sourceName":"New Atlas"},{"url":"https://www.geeky-gadgets.com/figure-ai-helix-robotic-model/","title":"Figure AI Helix vision-language-action model (7B comprehension + 80M control)","sourceName":"Geeky Gadgets"},{"url":"https://www.figure.ai/news/introducing-figure-03","title":"Introducing Figure 03, the general-purpose home humanoid (purpose-built for Helix) (Oct 2025)","sourceName":"Figure (official)"},{"url":"https://www.prnewswire.com/news-releases/figure-exceeds-1b-in-series-c-funding-at-39b-post-money-valuation-302556936.html","title":"Figure exceeds $1B Series C at $39B (Parkway, Brookfield, NVIDIA, Intel, LG, Salesforce, T-Mobile, Qualcomm) (Sept 2025)","sourceName":"Figure (PR Newswire)"},{"url":"https://www.figure.ai/news/series-c","title":"Figure Series C funds next-gen GPU infrastructure to power Helix's core perception/reasoning/control models (Sept 2025)","sourceName":"Figure (official)"}],"keyFacts":[{"label":"Verified commercial milestone","value":"May 13, 2026: Figure CEO demonstrated a team of Figure 03 robots running a full 8-hour autonomous shift on Helix-02, package sorting, no human intervention"},{"label":"Training data","value":"Learned from approximately 500 hours of demonstration data"},{"label":"Onboard runtime","value":"Embedded low-power onboard GPUs, no cloud"},{"label":"DEPLOY maturity rationale","value":"Verified 8-hour autonomous shift in commercial sortation; counted as commercial not production"}],"aliases":["Helix","Helix-02"],"collisionRisk":"high","reviewNote":null,"builtOnBrainId":null,"createdAt":"2026-05-28T14:06:06.969Z","updatedAt":"2026-05-30T21:47:16.470Z","jsonLd":{"@context":"https://schema.org","@type":"SoftwareApplication","@id":"https://registry.deploy.report/brains/helix","url":"https://registry.deploy.report/brains/helix","name":"Helix (and Helix-02)","alternateName":["Helix","Helix-02"],"description":"Figure AI's onboard dual-system VLA. System 2 is an internet-pretrained VLM (7B params, 7-9 Hz) handling scene and language understanding; System 1 is a fast reactive visuomotor policy (80M params, 200 Hz). Helix-02 (January 27, 2026) extended to full-body control via System 0, a 1 kHz neural prior trained on 1,000+ hours of human motion data that replaced roughly 109,504 lines of hand-engineered C++. The first VLA to control a full humanoid upper body including individual fingers from one set of weights, and the first to run two robots from one set of weights.","identifier":"9fc8ad4b-3d2f-4a3e-a0fe-f7ea7e138eb2","applicationCategory":"foundation-model","publisher":{"@id":"https://deploy.report/#organization"}},"framework_metadata":{"framework_schema_version":"0.1.0","verification_status":"verified","maturity_stage":"commercial","lifecycle_state":null,"architectural_position":{"cohort":null,"sub_cohorts":[]},"within_cohort_verified_vs_claimed_pair":null,"cap_flags":[],"verification_depth":{"sources_count":10,"primary_source_types":["primary-company-ir"]}}}