Caltech and Oratomic introduced mitten codes on August 3, 2026, a new family of non-abelian quantum low-density parity-check codes published on arXiv under the title "High-rate qLDPC processors." This is worth separating clearly from Oratomic's $300 million Series A, covered here recently: that post was about a funding round built on an earlier Caltech resource estimate. This is a different, later paper describing an actual code construction.
What a constant encoding rate buys you
Encoding rate is the ratio of logical qubits to physical qubits a code produces, and it's one of the two numbers (alongside check weight) that decide whether a qLDPC code is practical to build. Surface codes, the current default, have encoding rates that shrink as you scale up, which is exactly the overhead problem our logical qubits explainer covers in general terms. A code family with a constant rate, here reported at 20%, doesn't lose ground as the code grows larger. That property, combined with low check weight (few qubits involved per parity check, which matters for realistic hardware wiring), is what makes a qLDPC construction worth naming rather than another point in a resource-estimate table.
Non-abelian, and why that's the interesting part
Most qLDPC constructions getting attention recently, including the routing codes from USTC and Origin Quantum Computing covered here days ago, build on structures where the underlying group operations commute (abelian structures). Mitten codes reportedly use non-abelian group structures instead, a mathematically distinct approach to the same overhead problem. Two independent teams reaching for qLDPC code families from different corners of group theory in the same week is a real signal that this is where the field's attention on error-correction overhead currently sits, not evidence that either result is more correct than the other.
What's confirmed and what isn't
What's confirmed: the code family's name, its reported 20% constant encoding rate, low check weight, and the non-abelian construction, all as described in secondary reporting on the arXiv paper. What isn't confirmed here: independent verification of the rate claim, a hardware demonstration, or peer review. Same caveat as the routing codes coverage: this is a real, checkable theoretical contribution, not yet a result from working hardware.
What to watch next
Whether either Oratomic's mitten codes or USTC's routing codes gets picked up by a hardware team and fabricated is the test that matters. Our real-time decoding bottleneck piece covers the other half of this cost equation: a code with a great encoding rate still needs a decoder fast enough to keep up with it in practice.