Trapped-ion quantum computing startup IonQ today announced the launch of its latest quantum computer, which features what IonQ calls “32 perfect qubits with low gate errors.”

Using IBM’s preferred quantum benchmark, IonQ expects to hit a quantum volume of 4,000,000. That’s a massive increase over the double-digit quantum volume numbers that IBM itself recently announced and it’s a pretty extraordinary claim on IonQ’s side, as this would make its system the most powerful quantum computer yet.

The (well-funded) company has never used this metric before. Through a spokesperson, IonQ also noted that it doesn’t necessarily think quantum volume is the best metric, but since the rest of the industry is using it, it decided to release this number. The company argues that its ability to achieve 99.9% fidelity between qubits has allowed it to achieve this breakthrough.

“In a single generation of hardware, we went from 11 to 32 qubits, and more importantly, improved the fidelity required to use all 32 qubits,” said IonQ CEO and president Peter Chapman . “Depending on the application, customers will need somewhere between 80 and 150 very high-fidelity qubits and logic gates to see quantum advantage. Our goal is to double or more the number of qubits each year. With two new generations of hardware already in the works, companies not working with quantum now are at risk of falling behind.”

It’s worth noting that IonQ’s trapped-ion approach is quite different from IBM’s (or D-Wave’s for that matter) which uses a very different technique. That makes it hard to compare raw qubit counts between different vendors. The quantum volume metric is meant to make it easier to compare these systems, however.

“The new system we’re deploying today is able to do things no other quantum computer has been able to achieve, and even more importantly, we know how to continue making these systems much more powerful moving forward,” said IonQ co-founder and chief scientist Chris Monroe. “With our new IonQ system, we expect to be able to encode multiple qubits to tolerate errors, the Holy Grail for scaling quantum computers in the long haul.”

Using new error correction techniques, IonQ believes that it will only need 13 qubits to create a “near-perfect” logical qubit.

For now, IonQ’s new system will be available as a private beta, and it’ll be interesting to see if its early users will back up the company’s claims (unsurprisingly, given the magnitude of IonQ’s claims, there’s a bit of skepticism within the quantum computing community). Later, the company will make it available through partners like Amazon with its Braket service and the Microsoft Azure Quantum Cloud.