Real-world evaluations of physical AI
We measure and report frontier robotics capabilities to the public.
Opus 5 on block stacking
LLM inference speed over time
Test-time scaling in robotics capabilities
Opus 5 and the Tower of Hanoi
Opus 5 on block stacking
LLM inference speed over time
Test-time scaling in robotics capabilities
Opus 5 and the Tower of Hanoi
Backed by NVIDIA InceptionAnthropic for Open SourceFeatured by Google for Developers
Opus 5 and the Tower of Hanoi
Complete the Tower of Hanoi puzzle by moving all disks to the rightmost peg.
With support from experts at:
Independent analysis of robotics capabilities
Test-time scaling in robotics capabilities
Robotics capabilities of LLMs scale with thinking effort.
Source: Maccark, Opus 5 report (Aug 2026)Methodology
A technology this big must be measured in the open
Source: METR, Task-Completion Time Horizons of Frontier AI Models
The stakes
Robotics capabilities are progressing rapidly
Frontier labs are racing to develop general-purpose robots in the next two years.
The problem
No one knows where the frontier is
The Internet is filled with cherry-picked demo videos. No continuous, standardized evaluations for robotics exist.
Our answer
We are an independent evaluator for robotics
We measure robotics capabilities in the real world, independent of any agenda.
Our work
Open tools for open evaluation
Open source
Our tools and datasets are public so the community can verify and build on our work.
Contribute
About us
Maccark is a Public Benefit Corporation helping society understand the frontier of physical AI.
Our mission
Maccark is a Public Benefit Corporation bound by law to serve the public good. General-purpose robots may arrive within years, with profound implications for the economy and the labor market.
We build open-source tools and independent benchmarks to measure robotics capabilities. We report progress to the public rigorously and neutrally.
What experts say
Words from the experts
Researchers, engineers, and forecasters on Maccark.
"Supply chain automation timelines are a crucial input to ASI timelines in hardware-dependent AI takeoff scenarios, as well as forecasting progress in AI military technologies. This kind of benchmarking effort could help ground those forecasts."
Ryan KiddCEO & Co-Founder at MATS Research
"Timelines to robot automation is an important input into our models of takeoff, yet it is one that I (and in my experience, many other people) have a lot of uncertainty about. This project appears well-scoped to make progress on this."
Gabe WuAlignment Researcher at OpenAI
"There is a need for good robotics benchmarks to capture capability improvements that may emerge in the coming years. This project could mark a strong contribution to this space."
Julian JacobsResearch Scientist and Economist at Google DeepMind
"Supply chain automation timelines are a crucial input to ASI timelines in hardware-dependent AI takeoff scenarios, as well as forecasting progress in AI military technologies. This kind of benchmarking effort could help ground those forecasts."
Ryan KiddCEO & Co-Founder at MATS Research
"Timelines to robot automation is an important input into our models of takeoff, yet it is one that I (and in my experience, many other people) have a lot of uncertainty about. This project appears well-scoped to make progress on this."
Gabe WuAlignment Researcher at OpenAI
"There is a need for good robotics benchmarks to capture capability improvements that may emerge in the coming years. This project could mark a strong contribution to this space."
Julian JacobsResearch Scientist and Economist at Google DeepMind
"We've all seen the demo where a robot does a backflip or jumps rope, and then you try to use it for anything real and, in the best case, it doesn't break your own robot. That gap is exactly why independent measurement matters."
Liane GalantiPhD Student in Computer Science at Princeton
"Measuring robotics capabilities over time seems like a very important input for forecasting AI takeoff speeds. Currently, there are basically no widely-used high-quality robotics benchmarks, and additional high-quality evals here seem very valuable."
Nikola JurkovicMember of Technical Staff at METR
"Physical AI may be one of the most transformative technologies in human history, and yet we cannot say with any confidence what robots can do today, how fast it is changing, or when it will start to affect human labour markets."
Sebastian SartorPhD Student in Mechanical Engineering at MIT
"We've all seen the demo where a robot does a backflip or jumps rope, and then you try to use it for anything real and, in the best case, it doesn't break your own robot. That gap is exactly why independent measurement matters."
Liane GalantiPhD Student in Computer Science at Princeton
"Measuring robotics capabilities over time seems like a very important input for forecasting AI takeoff speeds. Currently, there are basically no widely-used high-quality robotics benchmarks, and additional high-quality evals here seem very valuable."
Nikola JurkovicMember of Technical Staff at METR
"Physical AI may be one of the most transformative technologies in human history, and yet we cannot say with any confidence what robots can do today, how fast it is changing, or when it will start to affect human labour markets."
Sebastian SartorPhD Student in Mechanical Engineering at MIT
