Qwen3.5 recipes shared for Jetson Thor

A Jetson Thor forum post shares setup recipes for running multiple Qwen3.5 models with NVIDIA's latest vllm repository for Thor. The largest reported working model is Qwen3.5-122B-A10B, with notes on NVFP4 and INT4 tradeoffs.

A new set of community recipes outlines how to run various models from the Qwen3.5 family on Jetson Thor. The shared guidance is based on NVIDIA’s latest vllm repository for Thor and is published through a linked GitHub repository intended for model deployment and experimentation.

The largest model reported to run is Qwen3.5-122B-A10B. The resharded NVFP4 version is described as a bit slower because it does not have MTP, but it is also described from experience as consistently better than the INT4 version. The note frames the recipes as practical instructions for users trying to evaluate performance and quality tradeoffs across quantized variants on Thor hardware.

The forum exchange is brief and focused on sharing working configurations rather than benchmarking details or broader documentation. A reply from an NVIDIA forum participant thanks the contributor for sharing the recipes, signaling community interest in reproducible ways to run large language models on Jetson Thor.

The surrounding forum activity shows sustained attention on generative model deployment on Thor, including discussions about Qwen variants, Nemotron, vllm containers, compatibility issues, and performance comparisons. That context positions the Qwen3.5 recipes as part of a larger effort by developers to tune local inference workflows for edge and robotics computing systems built on Jetson Thor.

55

Impact Score

Artificial Intelligence pushes practical change in claims

Claims operations are emerging as a key area where Artificial Intelligence is delivering practical gains through continuous monitoring, better decision-making, and reduced administrative burden. The shift is moving beyond automation toward changes in incentives, workflows, and the link between claims, underwriting, and pricing.

What SerDes does in high-speed chip communication

SerDes converts parallel and serial digital data for high-speed chip-to-chip links while reducing the number of interconnects required. It underpins physical layer connectivity across computing, automotive, mobile, and internet-connected systems.

LG Display starts mass production of 1-120 Hz laptop panel

LG Display has begun mass production of an LCD laptop panel using its Oxide 1 Hz technology. The panel dynamically shifts between low and high refresh rates based on onscreen activity to balance efficiency and responsiveness.

How the UK can strengthen regional climate tech clusters

Regional climate tech clusters across Scotland, the North East and the Midlands are gaining attention as growth engines beyond London. Targeted funding, better-connected accelerators and stronger Artificial Intelligence skills could help regional startups scale more effectively.

Oracle expands Artificial Intelligence database tools for business data

Oracle introduced new agentic Artificial Intelligence capabilities for Oracle Artificial Intelligence Database aimed at helping enterprises build, deploy, and secure production-grade applications on business data. The company is positioning the platform across multicloud, hybrid, and on-premises environments with open standards and stronger data controls.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.