Footer

Your go-to destination for discovering trending open-source projects and uncovering the insights that matter

Legal

  • Terms and conditions
  • Privacy policy

More

  • Pricing
  • Advertise
  • onedump
  • gosnakego
  • Contact us

Social

  • X (Twitter)
  • GitHub
Trendshift owl
What they build speaks louder than words
© 2026 Trendshift, created by Julian Li with ❤️
SignalAdvertise
Trendshift
SignalAdvertise
Sign in
Trendshift owl

Discover trends that matter

SignalAdvertise
DailyWeeklyMonthlyYearly
Live mentionsTopics
RepositoriesDevelopers
InsightsStats
Bifrost
Bifrost

giannisanni/pulsar

SSD-streaming inference engine for giant MoE models (Rust + CUDA). GLM 5.2 743B at 2 tok/s and Hy3 295B at 7 tok/s on two consumer 16GB GPUs. Zero-config multi-GPU: measures PCIe bandwidth, places attention and hot experts where they fit.

#LLM inference
Visit GitHub
Rust
234
30
1 contributors
Custom license
AchievementsView all →
FeaturedAdvertise to 180K+ monthly visitors
Bifrost

Fastest Enterprise AI Gateway for Routing, Governing, and Securing all your AI Traffic

Momentum

How fast giannisanni/pulsar is gaining stars, and where it lands on Trendshift and GitHub Trending

Activity and trending

Stars, forks and merged pull requests each day or month, lined up with the repository's rank on Trendshift Trending

Sep 8 – Nov 4, 2026

Stars +26 in 58 days

024
4

Forks +2 in 58 days

01
1

Merged PRs 0 in 58 days

01

Trendshift Trendingwhere the repository ranked each day it made the top 25

Not in the Trendshift top 25 in this period

Sep 8Sep 16Sep 24Oct 2Oct 10Oct 18Oct 26Nov 3
The left and right arrow keys step from day to day; Home and End go to the first and the last

Stars, forks, merged pull requests and Trendshift Trending rank, by day
DayStars gainedForks gainedMerged PRs
Tue, Sep 8, 2026+1+00
Wed, Sep 9, 2026+0+00
Thu, Sep 10, 2026+0+00
Fri, Sep 11, 2026+0+00
Sat, Sep 12, 2026+2+00
Sun, Sep 13, 2026+0+00
Mon, Sep 14, 2026+1+00
Tue, Sep 15, 2026+2+00
Wed, Sep 16, 2026+4+00
Thu, Sep 17, 2026+1+00
Fri, Sep 18, 2026+0+00
Sat, Sep 19, 2026+1+00
Sun, Sep 20, 2026+1+00
Mon, Sep 21, 2026+0+00
Tue, Sep 22, 2026+0+00
Wed, Sep 23, 2026+0+00
Thu, Sep 24, 2026+2+00
Fri, Sep 25, 2026+0+00
Sat, Sep 26, 2026+0+00
Sun, Sep 27, 2026+0+00
Mon, Sep 28, 2026+0+00
Tue, Sep 29, 2026+0+10
Wed, Sep 30, 2026+0+00
Thu, Oct 1, 2026+0+00
Fri, Oct 2, 2026+0+00
Sat, Oct 3, 2026+0+00
Sun, Oct 4, 2026+0+00
Mon, Oct 5, 2026+0+00
Tue, Oct 6, 2026+2+10
Wed, Oct 7, 2026+0+00
Thu, Oct 8, 2026+0+00
Fri, Oct 9, 2026+0+00
Sat, Oct 10, 2026+0+00
Sun, Oct 11, 2026+0+00
Mon, Oct 12, 2026+1+00
Tue, Oct 13, 2026+1+00
Wed, Oct 14, 2026+0+00
Thu, Oct 15, 2026+1+00
Fri, Oct 16, 2026+1+00
Sat, Oct 17, 2026+0+00
Sun, Oct 18, 2026+0+00
Mon, Oct 19, 2026+0+00
Tue, Oct 20, 2026+0+00
Wed, Oct 21, 2026+0+00
Thu, Oct 22, 2026+0+00
Fri, Oct 23, 2026+0+00
Sat, Oct 24, 2026+0+00
Sun, Oct 25, 2026+1+00
Mon, Oct 26, 2026+0+00
Tue, Oct 27, 2026+0+00
Wed, Oct 28, 2026+0+00
Thu, Oct 29, 2026+0+00
Fri, Oct 30, 2026+0+00
Sat, Oct 31, 2026+0+00
Sun, Nov 1, 2026+1+00
Mon, Nov 2, 2026+3+00
Tue, Nov 3, 2026+0+00
Wed, Nov 4, 2026+0+00
Total+26+20
  • 1#1 on the list
  • 7Rank that day
  • Not trending

Mentions and achievements

Where people are talking about giannisanni/pulsar, and every badge it has earned
  1. Aug 21 post

    • It a really interesting concept though, since a lot of LLMs nowadays are MoE, the actual active size is small, and since experts as a units are even smaller, how optimized can we make streaming…

      Read full on X, posted by @misaalanshori03

      @misaalanshori03 · x.com

  2. 23 days earlier
  3. Jul 101 post

    • Running Tencent Hy3, a 295B-parameter MoE, on one RTX 4060 Ti 16GB. How: NeutronStar (my CUDA fork of @antirez's ds4) keeps attention + shared experts resident and streams the routed experts off an…

      Read full on X, posted by @Giannisanii

      @Giannisanii · x.com

  4. Jul 41 post

    • this tweet sent me down a rabbit hole: 284B DeepSeek V4 Flash running on a single RTX 4060 Ti, experts streaming from a cheap NVMe via antirez's ds4. 6x speedup overnight + found and fixed the bug…

      Read full on X, posted by @Giannisanii

      @Giannisanii · x.com

You might also like

Repositories close to giannisanni/pulsar
FeaturedAdvertise to 180K+ monthly visitors
Bifrost

Fastest Enterprise AI Gateway for Routing, Governing, and Securing all your AI Traffic

siris9476/pulsarforge

C
235

pegainfer-project/pegainfer

Rust

Pure Rust + CUDA LLM inference engine — no PyTorch, OpenAI-compatible, serves Qwen3 to Kimi-K2

733126

Avarok-Cybersecurity/atlas

Rust

Pure Rust Inference Engine

713110

ferrumox/rabbit

Rust

Run Qwen 3.8 MAX (2.4T MoE) on a 50GB-RAM consumer machine in pure Rust, experts streamed from disk. Tiny engine, immense model.

19121

midagedev/bloomery

Rust

Hybrid GPU + CPU inference for mixture-of-experts models, written in Rust down to the CUDA kernels. A drop-in llama-server.

607

iamwavecut/ninfer-all

C++

One line of NInfer for the Blackwell, Ada Lovelace, Ampere, Turing and Volta NVIDIA GPUs, consolidated from the forks that carry it: GGUF block formats, ternary Bonsai 2, device route profiles, MTP and DFlash2.

18939

lateos-ai/reflex

Rust

A high-performance, GGUF-native Rust & CUDA inference engine optimized for cold-start latency and real-time 'System 1' agent decision loops.

651

tanishqkumar/ssd

Python

A lightweight inference engine supporting speculative speculative decoding (SSD).

1k82

naklecha/simple-llm

Python

~1300 line, minimal, extensible LLM inference engine built from scratch.

49238