What's New · Latest Quantum Milestones
A running timeline of the field's most consequential hardware and error-correction results. The Verified landmark results below are curated against primary or reputable secondary sources, with figures quoted exactly. The Live feed is an automated news stream — not a verified-milestone list. This page is refreshed during site builds; the last build date is shown in the footer.
Live feed · The Quantum Insider (news, not a verified-milestone list)
Pulled automatically from The Quantum Insider at each site build. These are news/industry headlines — reporting, funding, and announcements, not independently verified milestones. Headlines link out to the original reporting.
Verified landmark results
Curated, peer-reviewed milestones — figures checked against the primary source and cited at the foot of the page.
2026
These very recent 2026 results are summarized here as reported; treat the exact figures as provisional and defer to the official publications and vendor announcements for the final numbers.
Microsoft × Quantinuum — an 800× logical-vs-physical error gap, validated in Nature
Peer-reviewed results confirmed a roughly 800-fold reduction between physical and logical error rates on trapped-ion hardware (System Model H2, 56 fully connected qubits, quantum volume 225), encoding four logical qubits from thirty physical ions — the largest physical-to-logical gap independently validated to date.
Origin Quantum (本源) — Wukong-180, a 180-qubit single-chip superconducting QPU
China's fourth-generation superconducting machine went live on Origin's cloud with 180 computational qubits plus 251 coupling qubits on a single self-developed chip, reporting 99.9% single-qubit, 99% two-qubit, and 99% readout fidelity.
QuEra — 96 logical qubits from 448 neutral atoms
Using high-rate [[16,6,4]] codes operated below threshold, QuEra encoded 96 logical qubits from 448 physical atoms — the first demonstration of more than 48 verified logical qubits, roughly doubling the prior record and reported in Nature.
2025
IBM — Nighthawk and Loon
Nighthawk packs 120 qubits on a square lattice with 218 next-generation tunable couplers, designed to run circuits of up to 5,000 two-qubit gates (roadmapped to 7,500 by the end of 2026 and 10,000 in 2027). Loon, an experimental processor, demonstrated the full set of components for fault tolerance — c-couplers reaching beyond nearest neighbours and a qLDPC-oriented architecture — on IBM's path to a fault-tolerant system (Starling) by 2029.
IonQ — Tempo reaches #AQ 64
IonQ's Tempo trapped-ion system hit an algorithmic-qubit benchmark of #AQ 64 ahead of schedule, with on the order of 100 physical qubits.
USTC — Zuchongzhi 3.0
A 105-qubit superconducting processor from USTC demonstrated a random-circuit-sampling quantum advantage — a significant result alongside the Jiuzhang photonic line.
2024
Google — Willow and below-threshold error correction
On the 105-qubit Willow chip, Google showed that the logical error rate is suppressed by a factor of Λ ≈ 2.14 each time the surface-code distance grows by 2 (from 3 to 5 to 7) — error correction that improves with scale, published in Nature 638, 920 (2025).
The road ahead
Several major vendors have published roadmaps targeting fault-tolerant milestones around the early 2030s, but these remain roadmap targets, not guaranteed outcomes. A few anchored targets:
- IBM — IBM's roadmap targets practical quantum advantage demonstrations in 2026, a logical-memory module (Kookaburra) around 2026, and a fault-tolerant system (Starling, ~200 logical qubits) by 2029.
- IonQ — IonQ's roadmap targets a two-chip interconnected module with photonic networking (~20,000 physical qubits, on the order of 1,600 logical) by 2028.
- Atom Computing × Microsoft — "Magne," a roadmap target of 50 logical qubits from ~1,200 physical atoms, expected around the start of 2027.
Want to migrate these examples to a vendor's real-hardware workflow? Real-hardware / cloud submission is a planned capability — the reference build accepts vendor target names but runs locally on the CPU statevector simulator. See Prepare Your First Vendor Hardware Workflow and the per-vendor Hands-On guides.
References
- Microsoft and Quantinuum, "Improved quantum processor logical error rates via correction and detection," Nature 654, 349-355 (2026) — up to an 800-fold improvement in logical vs. physical error rate on trapped-ion hardware (System Model H2). doi:10.1038/s41586-026-10628-y · Microsoft Quantum blog
- Origin Quantum (本源量子), "Origin Wukong-180" — 4th-generation 180-qubit single-chip superconducting quantum computer, online May 2026. originqc.com.cn
- D. Bluvstein, A. A. Geim et al. (Harvard / QuEra Computing and collaborators), "A fault-tolerant neutral-atom architecture for universal quantum computation," Nature 649, 39-46 (2026) — universal fault-tolerant architecture using up to 448 neutral atoms, encoding up to 96 logical qubits with below-threshold error suppression. doi:10.1038/s41586-025-09848-5 · media coverage: IEEE Spectrum
- IBM, "IBM Delivers New Quantum Processors, Software, and Algorithm Breakthroughs on the Path to Advantage and Fault Tolerance" — Nighthawk (120 qubits, 218 tunable couplers) and Loon, 12 Nov 2025. newsroom.ibm.com
- IonQ, "IonQ Hits #AQ 64 Milestone Ahead of Schedule" — Tempo trapped-ion system, ~100 physical qubits (2025). ionq.com
- D. Gao, J.-W. Pan, X.-B. Zhu et al. (USTC), "Establishing a New Benchmark in Quantum Computational Advantage with 105-qubit Zuchongzhi 3.0 Processor," Phys. Rev. Lett. 134, 090601 (2025). arXiv:2412.11924 · doi:10.1103/PhysRevLett.134.090601
- Google Quantum AI, "Quantum error correction below the surface code threshold," Nature 638, 920–926 (2025). doi:10.1038/s41586-024-08449-y
最新进展 · 量子里程碑
下方「已验证里程碑」经过对照一手/可靠二手来源核实,数字照原文精确引用;「实时动态(Live feed)」是自动新闻流,并非已验证里程碑清单。本页在站点构建时刷新,最后构建日期见页脚。
实时资讯 · The Quantum Insider(新闻,非已核实里程碑列表)
每次构建站点时自动从 The Quantum Insider 拉取,周更追踪器保持其时效。标题均链接至原文报道。此处为新闻/行业头条(报道、融资、公告),并非经独立核实的里程碑。
重磅成果 · 已核实
精选、经同行评审的里程碑——数据均对照原始出处核实,文献见页面底部。
2026 年
以下 2026 年成果系据报道整理,具体数字以官方正式发布与厂商公告为准。
微软 × Quantinuum——逻辑与物理错误率差距达 800 倍,登上《自然》
同行评审结果证实,在离子阱硬件(System Model H2,56 个全连接量子比特,量子体积 225)上,物理与逻辑错误率之间约有 800 倍的差距,用 30 个物理离子编码出 4 个逻辑量子比特——这是迄今独立验证过的最大物理-逻辑差距。
本源量子——悟空 180,单芯片 180 比特超导量子计算机
中国第四代超导量子计算机在本源云上线,单颗自研芯片集成 180 个计算量子比特与 251 个耦合量子比特,实测单比特门保真度 99.9%、两比特门 99%、读出 99%。
QuEra——448 个中性原子编码出 96 个逻辑量子比特
采用阈值之下运行的高码率 [[16,6,4]] 码,QuEra 用 448 个物理原子编码出 96 个逻辑量子比特——首次实现超过 48 个经验证的逻辑量子比特,约为此前纪录的两倍,成果发表于《自然》。
2025 年
IBM——Nighthawk 与 Loon
Nighthawk 在方形晶格上集成 120 个量子比特与 218 个新一代可调耦合器,可运行多达 5,000 个两比特门的线路(路线图为 2026 年底 7,500 个、2027 年 10,000 个)。实验型处理器 Loon 演示了走向容错所需的全套组件——跨越近邻的 c 型耦合器与面向 qLDPC 的架构——这是 IBM 通往 2029 年容错系统(Starling)的关键一步。
IonQ——Tempo 达到 #AQ 64
IonQ 的 Tempo 离子阱系统提前达成算法量子比特基准 #AQ 64,物理量子比特数约为 100。
中科大——祖冲之三号
中科大的 105 比特超导处理器演示了随机线路采样的量子优越性,与九章光量子路线一道,是随机线路采样量子优势实验中的重要进展。
2024 年
谷歌——Willow 与阈值之下的纠错
在 105 比特的 Willow 芯片上,谷歌证明:表面码距离每增加 2(从 3 到 5 再到 7),逻辑错误率就被抑制约 Λ ≈ 2.14 倍——纠错能力随规模提升,成果发表于《自然》638, 920(2025)。
未来路线
多家主要厂商发布了面向 2030 年前后容错里程碑的路线图,但这些仍是路线图目标,而非确定结果。几个有明确锚点的目标:
- IBM——IBM 路线图目标是在 2026 年展示实用量子优势相关能力,2026 年前后推出逻辑存储模块(Kookaburra),2029 年推出容错系统(Starling,约 200 个逻辑量子比特)。
- IonQ——IonQ 路线图目标是在 2028 年前后实现带光子互联的双芯片模块(约 20,000 个物理量子比特、约 1,600 个逻辑量子比特)。
- Atom Computing × 微软——「Magne」,路线图目标是用约 1,200 个物理原子编码 50 个逻辑量子比特,预计 2027 年初投入运行。
想了解如何把示例迁移到厂商真机流程?向真机/云端提交属规划中能力——参考实现会接受厂商目标名称,但实际在本地 CPU 态矢量模拟器上运行。详见硬件任务准备指南,以及各厂商的SDK 指南。
参考文献
- Microsoft and Quantinuum, "Improved quantum processor logical error rates via correction and detection," Nature 654, 349-355 (2026) — up to an 800-fold improvement in logical vs. physical error rate on trapped-ion hardware (System Model H2). doi:10.1038/s41586-026-10628-y · Microsoft Quantum blog
- Origin Quantum (本源量子), "Origin Wukong-180" — 4th-generation 180-qubit single-chip superconducting quantum computer, online May 2026. originqc.com.cn
- D. Bluvstein, A. A. Geim et al. (Harvard / QuEra Computing and collaborators), "A fault-tolerant neutral-atom architecture for universal quantum computation," Nature 649, 39-46 (2026) — universal fault-tolerant architecture using up to 448 neutral atoms, encoding up to 96 logical qubits with below-threshold error suppression. doi:10.1038/s41586-025-09848-5 · media coverage: IEEE Spectrum
- IBM, "IBM Delivers New Quantum Processors, Software, and Algorithm Breakthroughs on the Path to Advantage and Fault Tolerance" — Nighthawk (120 qubits, 218 tunable couplers) and Loon, 12 Nov 2025. newsroom.ibm.com
- IonQ, "IonQ Hits #AQ 64 Milestone Ahead of Schedule" — Tempo trapped-ion system, ~100 physical qubits (2025). ionq.com
- D. Gao, J.-W. Pan, X.-B. Zhu et al. (USTC), "Establishing a New Benchmark in Quantum Computational Advantage with 105-qubit Zuchongzhi 3.0 Processor," Phys. Rev. Lett. 134, 090601 (2025). arXiv:2412.11924 · doi:10.1103/PhysRevLett.134.090601
- Google Quantum AI, "Quantum error correction below the surface code threshold," Nature 638, 920–926 (2025). doi:10.1038/s41586-024-08449-y