Specifications
Sourced specs
| Spec | Value | Source |
|---|---|---|
| Memory | 144 GB | via alibabacloud.com ↗2026-05-20 |
| Interconnect | 800 GB/s (chip-to-chip) | via alibabacloud.com ↗2026-05-20 |
| Precision support | Native FP32 down to FP4 | via alibabacloud.com ↗2026-05-20 |
| Performance | 3x predecessor Zhenwu 810E | via alibabacloud.com ↗2026-05-20 |
| Companion switch chip | ICN Switch 1.0: 25.6 Tbps aggregate bandwidth, congestion-free clusters of 64 accelerators | via alibabacloud.com ↗2026-05-20 |
| System availability | Panjiu AL128 Supernode Server: 128 accelerators per unit, PB/s internal bandwidth, offered via Bailian (China market) | via alibabacloud.com ↗2026-05-20 |
| Installed base (Zhenwu line) | More than 560,000 Zhenwu chips shipped; 400+ external customers across 20 industries | via alibabacloud.com ↗2026-05-20 |
| Software stack | T-Head SAIL | via alibabacloud.com ↗2026-05-20 |
Pricing
Price history, newest first
No public price observation yet.
vendor-official or multiple independent reports · single major report or several secondary · estimate or lone secondary source
Prices in USD, converted at each observation year's average exchange rate; the original figure and rate print under every converted price. How prices are sourced and graded ↗
Context
Notes
- Process node, compute TFLOPS figures, and price have not been disclosed; no TDP published.
- Compute figures (TFLOPS/TOPS) were not published in the announcement materials located.
- Unveiled alongside the Qwen3.7-Max flagship model as part of Alibaba Cloud's full-stack agentic-AI refresh.
Where it sits
In the stack
All specs and prices link to their original sources, dated as observed. Figures are market observations, not offers; treat each as of its date.