OpenRoboto

Robot intelligence built in the open. Bittensor subnet 80.

[OpenRoboto @openroboto]

Today, we’re officially launching OpenRoboto — Bittensor Subnet 80.

We’re sharing the full vision: open weights, open data and open benchmarks, coordinated through Bittensor to continuously improve robotics models.

[OpenRoboto @openroboto]

Glad to be backing Exploit Summit.

If you are in Montréal Sep 28-29 and care about where robotics intelligence gets built, come find us.

[Exploit Summit @ExploitSummit]

AI learned to write, draw, and code. The physical world is the one it's still struggling to crack.

@openroboto is going after it in the open as SN80. Open weights, open data, open benchmarks. Fine-tune the base robotics model, submit on-chain, and the best one becomes everyone's new starting point.

Closed labs hoard their robotics models. This one gets better in public, every single cycle.

A silver sponsor of Exploit 🤖https://t.co/CgJmJoRXKk

[Exploit Summit @ExploitSummit]

AI learned to write, draw, and code. The physical world is the one it's still struggling to crack.

@openroboto is going after it in the open as SN80. Open weights, open data, open benchmarks. Fine-tune the base robotics model, submit on-chain, and the best one becomes everyone's new starting point.

Closed labs hoard their robotics models. This one gets better in public, every single cycle.

A silver sponsor of Exploit 🤖https://t.co/CgJmJoRXKk

[OpenRoboto @openroboto]

Thanks for featuring us!

[Openτensor Foundaτion @opentensor]

https://t.co/HUeOtrAu2K

[Axis Robotics @axisrobotics]

Announcing our partnership with @OpenRoboto as Axis Robotics expands its data capabilities into the @opentensor ecosystem

OpenRoboto is an open competition on Bittensor (Subnet 80) for continuously improving robotics models. Miners fine-tune an open base model starting from π0.5, every submission is scored in randomized LIBERO-Pro environments, and only a strictly better model becomes the new base.

As the core data engine behind OpenRoboto, Axis is supplying over 3 million multi-modal trajectories into OpenRoboto’s Open Data Pool to accelerate physical AI model training for miners across the Bittensor network. Furthermore, Axis will support OpenRoboto's authoritative benchmarking platform with its Data-to-Model Pipeline, enabling robust, auditable evaluation for both data and models across multiple benchmarks.

This partnership establishes a robust, end-to-end framework uniting distributed data collection with verifiable model evaluation.

Learn more about OpenRoboto: https://t.co/9viK2AXyfy

[OpenRoboto @openroboto]

@Apex_SN1 @EdwardMcDonne12 Let’s cook 🧑‍🍳🧑‍🍳

[Mark Jeffrey @markjeffrey]

Bittensor brings 'worldwide contest-o-nomics' to robot AI development.

Which, hilariously, may mean humanoid AND other robotic efforts may soon find $TAO powering their leading edges.

[OpenRoboto @openroboto]

@daanisharif @axisrobotics We love human

[OpenRoboto @openroboto]

Open weights were never the hard part. Open data is.

The Open Data Pool is live, with Axis Robotics @axisrobotics as launching partner.

https://t.co/4ynIBhHcbt

[OpenRoboto @openroboto]

🤝🤝

[Axis Robotics @axisrobotics]

Decentralized robotics intelligence needs open data infrastructure to scale.

Congrats to @OpenRoboto on launching the Open Data Pool on @opentensor. As OpenRoboto enables open data access and model evaluation, @axisrobotics is powering the pipeline beneath with our compounding data engine.

More soon 🤝

[Axis Robotics @axisrobotics]

Decentralized robotics intelligence needs open data infrastructure to scale.

Congrats to @OpenRoboto on launching the Open Data Pool on @opentensor. As OpenRoboto enables open data access and model evaluation, @axisrobotics is powering the pipeline beneath with our compounding data engine.

More soon 🤝

[OpenRoboto @openroboto]

@axisrobotics @opentensor Excited for what's coming next!

[OpenRoboto @openroboto]

This is not just about better fine-tuning. If the pool gets big enough, we can pretrain our own base model from scratch.

That will take years, and we are not claiming it today.

Docs and quickstart →
https://t.co/4ynIBhHcbt

[OpenRoboto @openroboto]

Axis collects this data without expensive hardware. Contributors teleoperate simulated arms straight from a browser, and capture real-world manipulation on a phone using hand pose tracking.

[OpenRoboto @openroboto]

@cypherweb024 Thanks for the detailed introduction!

[OpenRoboto @openroboto]

@arizona_luxury @physical_int Stay tuned!

[OpenRoboto @openroboto]

We use π0.5, developed by @physical_int, as the base model for the OpenRoboto competition.

It’s one of the strongest open VLA base models for robot manipulation today.

Training a robotics base model from scratch requires enormous amounts of diverse data. Rather than starting from zero, miners build on π0.5 and compete to improve it.

This gives every miner a strong shared model to start from.

See how it works: https://t.co/qQ67Rv5LQo

[OpenRoboto @openroboto]

@ShizzyUnchained Thanks for the mention!

[OpenRoboto @openroboto]

Thanks for featuring us! Cooking now.

[Openτensor Foundaτion @opentensor]

https://t.co/BhZToz8eS8

[OpenRoboto @openroboto]

The website now includes Brand Assets, mobile navigation, and a more complete Queue result view.

[OpenRoboto @openroboto]

Over the past week, OpenRoboto shipped updates across the website, submission system, and benchmark evaluation pipeline.

Miners can now follow a submission from queue entry through live evaluation progress, final scoring, or rejection. Check out our tech update for the past week:

[OpenRoboto @openroboto]

We also updated the model submission rules and documentation:
- Submit a complete merged checkpoint
- Bare LoRA adapters will not pass pre-check
- Use `https://t.co/4MJ9qvfzlA submit` to run upload, burn, and announce consecutively
- Check Scan Rejections to diagnose failed submissions
- The 0.01 challenger margin applies to the overall average score