PulseAugur
EN
LIVE 10:02:03

Developer embeds GPU into LLM weights to run CUDA kernels

By PulseAugur Editorial · [2 sources] ·

A developer has created qwenGPU, an experimental model that integrates a SIMT GPU directly into the weights of Qwen3.5-2B. This novel approach allows the model to execute real CUDA kernels within its own forward passes, with 45 additional untrained layers built using Qwen's existing Gated DeltaNet and SwiGLU blocks. The developer reports minimal text drift from the base model and claims high accuracy, with the model matching a real NVIDIA Tesla T4 on several kernels. AI

IMPACT This research could lead to more efficient and integrated AI hardware-software co-design, potentially impacting how future AI models are trained and deployed.

RANK_REASON The item describes a novel research approach to integrating GPU computation directly into LLM weights, rather than a formal product release or frontier model announcement.

Read on Mastodon — mastodon.social →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

Developer embeds GPU into LLM weights to run CUDA kernels

How we ranked this

Signal score
1 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
The item describes a novel research approach to integrating GPU computation directly into LLM weights, rather than a formal product release or frontier model announcement.
Source corroboration
2 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
model release, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
1 days old
Coverage has settled into its steady-state source set.

Full methodology in our editorial standards.

COVERAGE [2]

  1. MarkTechPost TIER_1 English(EN) · Michal Sutter ·

    Meet qwenGPU: Qwen3.5-2B With 45 Compiled GPU Layers That Run CUDA

    <p>qwenGPU compiles a 4-lane SIMT GPU into Qwen3.5-2B as 45 untrained layers. The model writes every step of real CUDA launches itself, using Gated DeltaNet state as registers and memory.</p> <p>The post <a href="https://www.marktechpost.com/2026/11/01/qwengpu-simt-gpu-compiled-i…

  2. Mastodon — mastodon.social TIER_1 English(EN) · [email protected] ·

    A developer has compiled a SIMT GPU directly into the weights of Qwen3.5-2B, so the model runs real CUDA kernels inside its own forward passes. The 45 added lay

    A developer has compiled a SIMT GPU directly into the weights of Qwen3.5-2B, so the model runs real CUDA kernels inside its own forward passes. The 45 added layers are untrained and built from Qwen's own Gated DeltaNet and SwiGLU blocks, and text drift from the base model stays t…

    LINKS marktechpost.com/…/qwengpu-simt-gpu-compi…

RELATED ENTITIES

RELATED TOPICS