Huawei adds day-zero Ascend inference support for Kimi K3.1 Pro
Overview
Huawei's Ascend computing team says it completed day-zero inference adaptation for Kimi K3.1 Pro, the 2.8-trillion-parameter model Moonshot AI released on November 4.
Developers can deploy it through vLLM Ascend and SGLang on Atlas 800 A3 servers, Atlas 900 A3 SuperPoD and Ascend 950 SuperPoD series.
The vLLM Ascend documentation lists Kimi K3.1 Pro as experimental support in version 0.30.0. Atlas 800I A3 servers run a W4A8 quantized checkpoint, and 950DT products use a W4A8 MXFP variant. The guide says Atlas A2 and 950PR products are not yet supported.
Written by AI from the articles below · updated Nov 4, 9:37 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- PandailyNewsHuawei adds day-zero support for Kimi K3.1 Pro on Ascend systems
AIHuawei's Ascend computing team says it completed day-zero inference adaptation for Kimi K3.1 Pro, the 2.8-trillion-parameter model Moonshot AI released on November 4. Developers can deploy it through vLLM Ascend and SGLang on Atlas 800 A3 servers, Atlas 900 A3 SuperPoD and Ascend 950 SuperPoD series, with the vLLM Ascend documentation listing it as experimental support in version 0.30.0.
Heat Trend
Not enough continuous observations to show a trend yet.