Dream-1 Model Uses Video Prediction to Control Bimanual Robots for $3/Hour

Dream Labs has raised $90 million in seed funding to launch a public application programming interface for its universal robot foundation model, according to SiliconANGLE. This capital injection coincides with the release of Dream-1, a system designed to make robots more intelligent and capable of performing diverse tasks using simple video feeds.

The risk

The primary challenge for current industrial automation is a severe lack of general intelligence, despite the widespread presence of robotic hardware. Dream Labs co-founder and CEO Joel Jang notes that while over five million industrial robots operate globally in factories and warehouses, they are largely limited to single tasks. These machines cannot transfer acquired knowledge to other units or adapt to new jobs without extensive, specialized programming.

Who is exposed

Manufacturers and logistics companies investing in robotic hardware face significant operational friction due to the high cost of training. Although robotic arms are now affordable, selling for just a few thousand dollars, they ship without inherent intelligence. Buyers must spend months teaching each arm specific jobs, a process that must be repeated from scratch whenever new hardware is introduced.

Cost and compliance

The funding round was jointly led by B Capital and Kindred Ventures, with participation from A.Capital, Unusual Ventures, Wing VC, SV Angel, Scribble VC, and angel investor Adrian Aoun. Dream Labs is emerging from stealth with Dream-1, which it prices at $3 per hour. The model is designed to control bimanual robots it has never seen, on tasks it was never shown, in locations it has never visited.

What should you do?

Teams evaluating robotic automation should assess whether their current hardware requires custom programming for every new task. Organizations should monitor the Dream Labs API launch to determine if video-prediction models can reduce their training timelines. Managers should compare the $3 hourly rate against the internal costs of specialized programming.

The background in plain language

Foundation models are large artificial intelligence systems trained on vast data sets to perform a wide range of tasks, rather than being programmed for one specific job. In robotics, this means creating a "universal brain" that can generalize skills across different machines and environments. Video prediction involves using AI to anticipate future frames in a sequence, allowing a robot to understand physical actions and outcomes without explicit step-by-step coding for every movement.

What to do and how to stay safe: Dream Labs

• Review current robotic workflows to identify tasks that require repetitive, manual programming for each new unit.
• Evaluate whether existing hardware supports external API integration for foundation model control.
• Monitor vendor announcements for updates on how these models handle unseen environments and tasks.
• Assess the total cost of ownership, including training time, against subscription-based AI control services.

Step-by-step guide: Browser security settings are the architectural walls of your digital life

General security guidance from the Vector Tribune newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

How much funding did Dream Labs raise?

Dream Labs raised $90 million in seed funding, according to SiliconANGLE.

What is the cost of using the Dream-1 model?

The service costs $3 per hour to control bimanual robots.

Who led the investment round?

The round was jointly led by B Capital and Kindred Ventures, with participation from several other investors.