In June, XDOF released ABC-130K: 3,553 hours of bimanual manipulation, 134,806 episodes, 195 tasks, free under Apache 2.0. It is the largest open dataset on the I2RT YAM arm, which is the arm a growing share of manipulation teams have standardised on. Two months earlier the Allen Institute open-sourced MolmoAct2 along with the 720 hours it was trained on, same arm, weights included.
That raises an obvious question. If thousands of hours of high-quality data on your exact hardware are free, why is anyone still paying to collect their own?
Many teams still do. A team I looked at this week has eleven people and is six months old, and it is publishing their own YAM teleoperation episodes one batch at a time while also fine-tuning MolmoAct2. They already have the free data, and they are collecting their own anyway.
That is not irrational, and the reason is the most useful thing I have learned about robot data collection this year.
One scoping note before going further, because the answer is different depending on who is asking. If you are one of the handful of labs training a robot foundation model from scratch, free data at this scale is genuinely valuable and you should take all of it. Everything below is aimed at the other group: the far larger number of teams building on top of someone else's base model, fine-tuning and post-training it for a specific robot doing a specific job. That is where the free data stops being enough, and that is most of the market.
What robot data is actually free now?
The floor moved this summer, and it moved a long way. A year ago, "we have bimanual manipulation data" was a defensible position for a startup, and today anyone can download it.
| Dataset | Hours | Episodes | Arm | Access |
|---|---|---|---|---|
| XDOF ABC-130K | 3,553 | 134,806 | I2RT YAM | Free, gated |
| Ai2 MolmoAct2-BimanualYAM | 720 | 34,500 | I2RT YAM | Free, open |
| Ai2 MolmoAct2-DROID | not stated | 74,604 | Franka | Free, open |
Add the open weights trained on top of them and you can go from nothing to a working manipulation policy on a common arm without collecting a single episode or signing a single contract. That is genuinely new, and anyone selling robot data who has not internalised it is selling last year's product.
Why is the same arm not the same setup?
This is where the free data stops being enough.
An episode is more than an arm trajectory, because the trajectory depends on a camera at a particular height and angle, a specific gripper, a table at a specific distance, particular lighting, and the specific objects in front of the robot. Change any of those and the policy that was trained on it degrades, quietly and unevenly.
You can watch this happen in public. On the ABC-130K discussion page, one researcher asks which sessions use the FlexPoint gripper rather than the standard one. Another asks for the extrinsic calibration for the wrist cameras. These read less like complaints and more like someone trying to use 3,553 hours and discovering that you cannot condition on what is not labelled.
That is the difference between a dataset you can download and a dataset you can deploy on. The hours are real, but the metadata that would make them useful for your setup is often missing.
What do the numbers say?
The strongest public evidence I have seen on this came out of an evaluation study by Parametric, a team that runs YAM arms with their own GELLO leader-arm rig. They compared training on data collected on the same embodiment against data from a similar but different one, on a hard deformable task.
| Training data | Success rate |
|---|---|
| Same embodiment as the robot being deployed | 77% |
| Cross-embodiment, similar arm | 7% |
That is an elevenfold difference, not an eleven percent one. On a task where contact and deformation matter, data from a nearly identical arm was worth almost nothing.
That number deserves some caution, since it comes from one study, one task class, and deformables are the hardest case. But the direction has been corroborated everywhere I have looked this year, and it inverts the pitch the whole data industry has been making. "Our hours will transfer to your robot" is a claim that now has a published counter-example.
So what should you actually collect?
It is tempting to reach for the language model analogy here, so let me be careful with it, because the loose version of it is wrong.
Pretraining did not become a commodity in language. OpenAI, Anthropic and Google all still pretrain their own frontier models, at enormous cost, and that pretraining run is arguably their central asset. What changed is that almost nobody else has to. For the thousands of companies building on top, the base model became something you rent or download, and the money moved to the layer above it: instruction tuning, preference data, evaluation, retrieval over their own corpus.
So a more honest framing is that pretraining is trending toward commoditisation for everyone who is not building the base model themselves.
Robotics is early in the same split, and the split matters more than the trend.
If you are building a robot foundation model, general data is still your core asset. You need vastly more of it than exists publicly today, ABC-130K and MolmoAct2 are a useful floor rather than a substitute, and every additional open corpus genuinely helps you. There are maybe a dozen organisations worldwide in this position.
If you are deploying a policy on top of someone else's base model, the calculus inverts. The base is increasingly something you download, built with free hours, open weights and someone else's capital. Trying to sell data at that layer means competing with a Hugging Face link.
The asymmetry between those two groups is the part people skip. Foundation model builders are a handful of well-funded labs. Teams fine-tuning and post-training an existing model for a specific robot, a specific cell, a specific customer, number in the hundreds and are growing much faster. Almost everyone reading this is in the second group, and almost every dollar of robot data spend will eventually come from it.
For that second group, and only that group, the expensive part is what comes after: a few hundred hours of your task, in your scene, with your gripper and your camera rig. That data is small in volume, but it is where the reliability actually comes from.
This is why an eleven-person team with free data on their shelf still runs teleoperation sessions. They are not trying to replace ABC-130K, because they need something that ABC-130K cannot give them.
The budgeting consequence is the same either way, and it is uncomfortable for anyone selling volume: the most valuable data is usually the most specific data, not the biggest dataset.
What should you ask if you are buying robot data?
Three questions worth asking any robot data vendor:
- What is in this that I cannot download? If the answer is hours on a common arm doing common tasks, the answer is nothing.
- What is the metadata? Gripper variant, camera extrinsics, lighting, object set, per-episode. Unlabelled hours are hours you cannot condition on, which makes them pretraining filler.
- Does it match my embodiment, or does it merely resemble it? 77 against 7 is the price of getting that wrong.
The free data is real and you should take it. If you are training a base model it may be the most valuable thing you download this year. If you are building on top of one, it improves performance up to a point. What it will not do is carry you to a deployment-ready standard, and the gap between those two things is wider than the download makes it look.
Related: The Data Wall on why demonstration data is the bottleneck at all, and How Robots Learn From Us on where these demonstrations come from in the first place.