CortexAI

Building the world's most diverse real-world, real-workplace, and industry-scale egocentric + robot dataset.

[CortexAI @cortexairobot]

Introducing Cortex Harness, a platform for running AI models on real robots with remote human guidance.

Prompt a robot to act. Steer it with language. Let a human take over remotely when needed to keep tasks moving.

Choose from GPT-6 Astra (Ultrafast), Claude, Gemini, open-weight models, and VLAs.

Here’s how it works 🧵

[Dream Labs @dreamlabs_ai]

@robocurve Anywhere you have an internet connection, you can prompt Dream-1 to control your robot.

Here are rollouts from beta testers from 8 countries, and from our office in SF.

[CortexAI @cortexairobot]

@dreamlabs_ai Congrats @dreamlabs_ai team! 🎉

[CortexAI @cortexairobot]

@kimmonismus Thanks Chubby! Excited to get more of them doing real work 🙌

[CortexAI @cortexairobot]

Introducing Cortex Harness, a platform for running AI models on real robots with remote human guidance.

Prompt a robot to act. Steer it with language. Let a human take over remotely when needed to keep tasks moving.

Choose from GPT-6 Astra (Ultrafast), Claude, Gemini, open-weight models, and VLAs.

Here’s how it works 🧵

[CortexAI @cortexairobot]

@Just_sharon7 Yes that's the goal! Keep robots doing useful work today, and use recovery data to reduce human intervention over time 🔥

[CortexAI @cortexairobot]

@SuveenE Let's go! 🦾

[CortexAI @cortexairobot]

@vishalsingh2972 That's the goal! Keep robots doing useful work today, and use recovery data to reduce human intervention over time 🔥

[CortexAI @cortexairobot]

@CyberRobooo Exactly! Cortex Harness connects AI models to real robots, with humans able to step in remotely when needed. Each recovery becomes data for future training.

[CortexAI @cortexairobot]

@Scobleizer Thanks Robert! A human in Singapore helping a robot 8,000 miles away in SF recover with minimal latency, then handing control back to the model. That’s what we're building toward: robots that keep doing useful work, with less human intervention over time.

[CortexAI @cortexairobot]

We're also inviting researchers, developers, and industry partners to our beta program. Participants will receive $100 in credits.

Read more about Cortex Harness and join our waitlist here 📚

https://t.co/mDSUnx8E3W

[CortexAI @cortexairobot]

4. Capture every recovery 🎞️

Each intervention records where the model got stuck, how a human helped, and what happened next so models can learn from every recovery.

The goal is a continuous improvement loop: deploy, identify failures, recover with human help, collect recovery examples, train, and redeploy.

[Boris Power @BorisMPower]

This confirms what Twitter demos made pretty clear in the first few days. Great to see an actual eval!

[CortexAI @cortexairobot]

To train MolmoAct 2 in collaboration with @allen_ai, we created the MolmoAct 2-Bimanual YAM dataset, a 700-hour collection of robot demonstrations involving two machine arms working together.

You can learn more about it here. 📊

https://t.co/6jtJF8WD93

[CortexAI @cortexairobot]

We tested GPT-6 Astra, Gemini 3.8 Flash, and Fable 5.1 on two real-world robot tasks from our MolmoAct2 real-world evaluations:
1. Fitting a 5 mm pipette tip
2. Loading a microwave

Astra scored highest on both. 🥇

Here are some few surprising things that Astra did to come out on top. 🧵

[CortexAI @cortexairobot]

You can see the full results and videos in our blog post here. 📝

https://t.co/DdnLnwdFGR