Figure's Helix 2.5 Hits 56% Success in Unfamiliar Homes: what are the key facts?

Figure released the Helix 2.5 system, with robots performing tasks such as tidying toys, folding towels, and making beds in 30 unfamiliar homes.

Figure's Helix 2.5 Hits 56% Success in Unfamiliar Homes: what are the key facts?

Test results show a 56% success rate on complete tasks; the figure is based on internal company evaluation and counts any human intervention as a failure.

Figure's Helix 2.5 Hits 56% Success in Unfamiliar Homes: what are the key facts?

After introducing Index pretraining, task success rate in controlled experiments rose from 9% to 56%, significantly improving cross-scenario generalization.

Figure's Helix 2.5 Hits 56% Success in Unfamiliar Homes: what are the key facts?

Figure previously accumulated 1,250 hours of operation at a BMW factory, handling over 90,000 parts, and is now shifting to home scenarios to validate generality.

Physix Frontier · News Briefing Card (enbrief · Sep 22, 2026)

Figure's Helix 2.5 Hits 56% Success in Unfamiliar Homes

KEY FACTS

Figure released the Helix 2.5 system, with robots performing tasks such as tidying toys, folding towels, and making beds in 30 unfamiliar homes.
Test results show a 56% success rate on complete tasks; the figure is based on internal company evaluation and counts any human intervention as a failure.
After introducing Index pretraining, task success rate in controlled experiments rose from 9% to 56%, significantly improving cross-scenario generalization.
Figure previously accumulated 1,250 hours of operation at a BMW factory, handling over 90,000 parts, and is now shifting to home scenarios to validate generality.

KEY DATA

56%Task success rate

30Number of test homes

$39 billionSeries C valuation

1,250 hoursBMW factory operating hours

PHYSIX OBSERVATION

Figure moves from structured factory floors to unstructured homes, validating the deployment potential of general-purpose embodied intelligence. Although the 56% success rate falls short of commercial standards, zero-shot generalization is a key signal for lowering deployment costs. Compared with Tesla's manufacturing advantage, Figure builds its moat through a self-developed model and large-scale human video data, seeking to prove its tech stack's adaptability in complex dynamic environments. This will be the watershed determining whether humanoid robots can truly enter the consumer market.

Source: enbrief original report ↗