NVIDIA/cutlass

CUDA Templates and Python DSLs for High-Performance Linear Algebra

Language modelsC++#cuda#deep-learning#deep-learning-library#cpp#nvidia

1.5 Steady Magnitude out of 10 — how fast interest is growing, not a quality score. Star growth looks organic. Data as of September 23, 2026, 12:53 UTC.

Open on Seismograph Open on GitHub

About the project

CUTLASS is a collection of CUDA C++ templates and Python DSLs (CuTe DSL) for implementing high-performance linear algebra, primarily GEMM, on NVIDIA GPUs. It targets researchers and performance engineers writing optimized Tensor Core kernels.

Useful for

Write a custom GEMM kernel in CUDA C++ with specific tiling and data types
Prototype a kernel in CuTe DSL using Python without deep C++ expertise
Run mixed-precision FP8/FP4 on Tensor Cores of Hopper or Blackwell architectures

README summarized by DeepSeek V4.1 Flash. Details may be inaccurate.

Why it’s trending

6 stars so far today, about 9 expected by the end of the day.
GitHub Trending C++ today: #15, +6 stars.
Recent forks include notable developers: @FindHao (325 followers).

Stars per day

Bars are daily stars, the line is the usual pace. Red marks spike days.

Numbers

Total stars: 10,474
Today: 6 · ≈ 9 by evening
Forks: 2,098
Issues and pull requests: 3,314
Watchers: 121
Language: C++
Latest release: v4.8.0 · September 22, 2026
Created: November 30, 2017
Last push: September 23, 2026

Star trust

Growth looks organic: forks and discussion are in line with active projects, and stars arrive unevenly, the way people give them.

These are heuristics, not a verdict: we judge by the repository’s behavior, not by a list of stargazers.

Spotted in

September 23, 2026GitHub Trending C++ today: #15, +6 stars

README badge

Paste it into your README — the badge shows the current magnitude and links to this page.

[![Seismograph](https://gitnova.dev/badge/NVIDIA/cutlass.svg?lang=en)](https://gitnova.dev/en/r/NVIDIA/cutlass)

More in this category

9.0
jaredpalmer/kev
kev is a LoRA adapter with a small readout head on top of Qwen2.5-0.5B that answers many typed questions about a document in a single forward pass, returning calibrated probabilities instead of text.

BreakoutLanguage modelsPython+464 stars today, ≈ 1,121 by evening
7.1
mizorewww/laya-mlx
Native MLX runtime for Laya typed decision models: returns probabilities for choices, scores or truth values without text generation, running locally on Apple Silicon.

PeakingLanguage modelsPython+134 stars today, ≈ 464 by evening
6.8
volotat/mini-AGI
A byte-level continual-learning language model that assembles its own architecture, trains from scratch on a single 8 GB VRAM GPU, and keeps learning from everything it reads. An experiment showing continual learning…

Early signalLanguage modelsPython+32 stars today, ≈ 108 by evening
6.4
TheoLeeCJ/SemIf
An open reproduction of the Jev-style semantic decision interface: reads typed option probabilities directly from a 4B model's logits without generating text. Runs on a single RTX 3090.

PeakingLanguage modelsPython+68 stars today, ≈ 211 by evening
6.2
nokia-applied-research/AnyJev
Turns any LLM into a decision model with typed questions (choice, yes/no, score) that return real probabilities read from the next-token distribution, without training. L0 removes position and label-prior bias; L1 adds…

Early signalLanguage modelsPython+122 stars today, ≈ 213 by evening
5.6
taeold/djev-run
Deploys the DiffusionGemma-Jev (djev) model on Google Cloud Run with an NVIDIA RTX PRO 6000 GPU and a TypeSafe AI compatible API. Includes Snake, Dino, and Tetris demos running directly in the browser.

Early signalLanguage modelsHTML+104 stars today, ≈ 190 by evening