← AI Switchboard
Robotics · Generalist AI · robots, stage, customers and money
← Robotics

Generalist AIPreview

Builds GEN-series AI models that control robot arms and hands, trained on half a million hours of real-world interaction data.

Leader Pete Florence, Co-founder and CEOBased in San Mateo, CaliforniaShares privateGroup Robot brains · the AI that runs robotsgeneralistai.comits news page

Where it stands: GEN-1 opened to early-access partners on 2 Apr 2026; no customers named. source

Venture-backed; June 2026 round led by Radical Ventures with NVIDIA, Bezos Expeditions and others participating. source

Models and platforms3 · newest first

Latest19 Aug 2026Robot model (AI)Research

GEN-1.5

Generalist's newest robot foundation model, a multimodal model that takes video, sensor, language and joint-position inputs and outputs robot actions. Its headline feature is one-shot learning: shown a single 3 to 12 second demonstration, it can attempt a new task with no retraining. Generalist has not named the robots or customers using it, or said who can access it.

  • 30 seconds of video memory
  • Outputs 100 Hz action trajectories
  • One-shot in-context learning from a single 3-12 second demonstration: 59% average success across 10 tasks
  • Few-shot: 83% average success after 10 gradient steps on 5 minutes of data per task
  • Pretrained continuously for over eight months
2 Apr 2026Robot model (AI)Preview

GEN-1

General-purpose robot model that outputs robot actions in real time. It was pretrained from scratch on half a million hours of real-world data, gathered with wearable devices worn by people rather than robots. Generalist says it averages 99% success on tasks where earlier models reached 64%, using about one hour of robot data per task. Opened to early-access partners; none named.

  • Pretraining data: half a million hours of real-world data, no robot data in pretraining
  • About 1 hour of robot data per reported task
  • 99% average success on tasks where previous models achieved 64%
  • Box assembly in 12.1 seconds; phone packing in 15.5 seconds
  • July 2026: works across about 9,000 end-effector variations, from grippers to tools
  • Access: early-access partners
4 Nov 2025Robot model (AI)Research

GEN-0

Generalist's first family of embodied foundation models, trained on over 270,000 hours of real-world manipulation data and using what it calls Harmonic Reasoning to think and act at the same time. Tested on 6- and 7-degree-of-freedom arms and 16+ DoF semi-humanoid robots, and shown publicly in a live box-packing demo on Universal Robots arms at NVIDIA GTC 2026.

  • Scaled to 10B+ parameter model sizes
  • Over 270,000 hours of real-world manipulation data, growing by over 10,000 hours per week
  • Robots tested: 6DoF, 7DoF and 16+DoF semi-humanoid

Timeline · customers, money, people, products

  • 25 Aug 2026MoneyRaised nearly $200M more, led by 8VC, extending its Series B to $600M at a $3B valuation, according to sources. Reported TechCrunch
  • 19 Aug 2026ProductReleased details of GEN-1.5, a robot model that can attempt a new task after a single 3-12 second demonstration. Generalist AI blog
  • 23 Jul 2026ProductSaid GEN-1 now works across about 9,000 end-effector variations, from five-finger hands to screwdrivers and peelers. Generalist AI blog
  • 4 Jun 2026MoneyRaised $400M led by Radical Ventures, with NVIDIA, Bezos Expeditions and others; total funding now more than half a billion dollars. Generalist AI blog
  • 1 Jun 2026PartnerNamed a founding member of NVIDIA's Cosmos Coalition for open world models, alongside Agile Robots, Skild AI and Runway. NVIDIA press release
  • 2 Apr 2026ProductLaunched GEN-1 to early-access partners, claiming 99% average success on tasks where earlier models reached 64%. Generalist AI blog
  • 24 Mar 2026PartnerReported its first public live demo: GEN-0 packing boxes at Universal Robots' GTC booth on UR7e arms with a MiR base. Generalist AI blog
  • 4 Nov 2025ProductIntroduced GEN-0, embodied foundation models trained on over 270,000 hours of real-world manipulation data. Generalist AI blog
← All companies← FieldAIGoogle DeepMind →