DEPLOYDatabase

Brain

GEN-1

Generalist AI's second-generation physical foundation model. Demonstrated 99% reliability on diverse tasks, execution up to 3x faster than prior state of the art, complex new physical skill learning, and emergent improvisational intelligence. Trained from scratch — goes beyond VLAs and world models. Supports broad range of end effectors.

Foundation model · Maturity: Pilot · Closed


Machine-readable surfaces

Architecture

Physical foundation model trained from scratch on real-world data at scale

Key facts

Mastery 99% claims (primary Generalist)
Generalist AI's April 2 2026 GEN-1 blog says GEN-1 improves average success rates to 99% on tasks where previous models achieve 64%, completes tasks roughly 3x faster than prior SOTA, and requires about 1 hour of robot data for those results. It defines mastery as reliability, speed, and improvisation. Treat benchmark comparisons as company-stated (Generalist AI)
Pretrain data scale (primary Generalist)
The same Generalist blog says GEN-1 is trained from scratch on a dataset that now includes over half a million hours of high-fidelity physical interaction data, with the base foundation trained without robot data (using wearable human interaction data) and task adaptation using roughly one hour of robot data. Treat dataset size as company-stated (Generalist AI)
Mastery threshold claims (primary Generalist)
Generalist AI (Apr 2 2026) introduced GEN-1 as an embodied foundation model it says crosses a mastery threshold on simple physical tasks, improving average success rates to 99% on tasks where prior models achieved 64%, completing tasks roughly 3x faster than prior SOTA, with about 1 hour of robot data per result. Treat performance claims as company-stated (Generalist AI)
GEN-1.5 one-shot learning
Generalist AI's follow-on GEN-1.5 post describes its embodied foundation-model research as targeting one-shot learning for physical tasks; this is the company's stated continuation of the GEN-1 line (Generalist GEN-1.5).
Beyond-world-models position
Generalist AI says its physical-intelligence approach goes beyond using a world model alone by connecting perception, reasoning, and action in an embodied system; this is a company position, not an independent evaluation (Generalist beyond world models).

Developed by (1)

Common questions

What is GEN-1?
Generalist AI's second-generation physical foundation model. Demonstrated 99% reliability on diverse tasks, execution up to 3x faster than prior state of the art, complex new physical skill learning, and emergent improvisational intelligence. Trained from scratch — goes beyond VLAs and world models. Supports broad range of end effectors.
Who developed GEN-1?
GEN-1 is credited to Generalist AI on the DEPLOY registry. Each developer attribution is verified via primary sources.
Is GEN-1 open source?
No. GEN-1 is recorded as proprietary on the DEPLOY registry. Model weights and source are not publicly available.
What type of AI is GEN-1?
GEN-1 is a foundation model, built on a Physical foundation model trained from scratch on real-world data at scale architecture on the DEPLOY registry.
What is GEN-1's maturity stage?
GEN-1 is at the pilot stage on the DEPLOY maturity ladder. Pilot stage means at least one named-customer trial deployment is verified.
Which robots run on GEN-1?
No robot models on the DEPLOY registry are recorded as running GEN-1. DEPLOY wires brain-to-model connections only when the wiring is verifiable from primary sources; absence may reflect pre-deployment or unverified manufacturer claims.

Sources (3)

  1. https://generalistai.com/blog/gen-1
  2. https://generalistai.com/blog/gen-1.5
  3. https://generalistai.com/blog/beyond-world-models
Methodology: Verified · 3 sources (no primary) · last reviewed 2026-10-09

Verification posture

Verified

Low confidence

Review state

Stable

Last reviewed 2026-10-09

Maturity + lifecycle

Maturity stage: pilot

Sources by quality tier

3
unclassified
Unclassified source

The framework is documented at /methodology. Corrections at /corrections. Reviewer: DEPLOY editorial team.

Methodology surface for GEN-1.

Canonical ID f08551f6-2cd5-436c-b483-87766bf3a316