twinspark

Production-grade recipe for DeepSeek-V4-Flash-0731 (284B MoE) on 2x NVIDIA DGX Spark: self-healing 2-node vLLM cluster, reboot-verified, tuned DSpark speculative decoding (~75 tok/s), full benchmarks, OpenAI Codex CLI integration

raullenchai/twinspark

View on GitHub ↗

Production-grade recipe for DeepSeek-V4-Flash-0731 (284B MoE) on 2x NVIDIA DGX Spark: self-healing 2-node vLLM cluster, reboot-verified, tuned DSpark speculative decoding (~75 tok/s), full benchmarks, OpenAI Codex CLI integration

twinspark grew -1.7% in the last 30 days, ranks 70th of 99 in Distributed Inference, last commit 100 days ago.

Star trajectory

83 days tracked
2026-08-30173 stars today2026-11-22
Momentum

-1.7% in 30d, below the Distributed Inference median of +1.2%.

Maintenance

Last commit 100 days ago, staler than the median of 5d.

How it compares in Distributed Inference

twinspark
173
stars
-1.7% 30d
commit 100 days ago
Distributed Inference median
299
stars
+1.2% 30d
commit ~5d ago
Leader · exo
48.5k
stars
+1.1% 30d
commit 15 days ago

Ranked 70th of 99 by stars.

Ranked by shared tags, then keywords, then stars
  1. RapidLLM is an LLM inference framework with continuous batching, tensor and data parallelism, quantized…

    +3 (+1.5%) in 30d · Committed 43d ago · Python

    Shares Local Runtime / Inference Engine, Inference Optimization, Distributed Inference

  2. Entrpi/ds4, a Blackwell CUDA perf fork of antirez/ds4 on NVIDIA DGX Spark: one-command install, ~3x upstream…

    +6 (+1.5%) in 30d · Committed 87d ago · Shell · MIT

    Shares Local Runtime / Inference Engine, Inference Optimization, Distributed Inference

  3. SSD-streaming inference engine for giant MoE models (Rust + CUDA). GLM 5.2 743B at 2 tok/s and Hy3 295B at 7…

    +9 (+3.9%) in 30d · Committed 81d ago · Rust · Other license

    Shares Local Runtime / Inference Engine, Inference Optimization, Distributed Inference

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

    +25 (+3.5%) in 30d · Committed yesterday · Rust · Apache-2.0

    Shares Inference Optimization, Distributed Inference, Local Runtime / Inference Engine

  5. Kindi 0.6.1: fast local LLM inference on low-cost GPUs and CPUs. DeepSeek-V4.1-Flash (552B) at 18.7 tok/s on…

    Committed 5d ago · Python · AGPL-3.0

    Shares Local Runtime / Inference Engine, Distributed Inference, Inference Optimization

  6. An agentic system that auto-optimizes LLM workloads on AMD GPUs.

    +33 (+13%) in 30d · Committed today · Python · Other license

    Shares Inference Optimization, Distributed Inference, Local Runtime / Inference Engine

  7. UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL…

    +39 (+2.5%) in 30d · Committed yesterday · C++ · Apache-2.0

    Shares Distributed Inference, Inference Optimization

  8. Local inference that behaves like a frontier API: Qwen3.8-Flash-Next on AMD Strix Halo plus small NPU models…

    +341 (+32%) in 30d · Committed 3d ago · Shell · Other license

    Shares Local Runtime / Inference Engine, Inference Optimization

  9. Zero-GPU CPU inference for 30B-284B MoE models (Qwen3.8 176.5B cold-cache certified in 28.26/48.35 GiB Cgroup…

    Committed today · Python · Apache-2.0

    Shares Local Runtime / Inference Engine, Inference Optimization

  10. Run large AI models on your own PC: Qwen3.8-Flash-Next on Strata's engine, and GLM 5.3 Flash, DeepSeek V4…

    +962 (+243%) in 30d · Committed 3d ago · C++ · MIT

    Shares Local Runtime / Inference Engine, Inference Optimization

Language
Shell
License
MIT License
Contributors
1
Forks
20
Watchers
17
Open issues
0
Releases
0
Created
2026-08-11
Stars this week
-3

You've seen where twinspark stands today.

Get the moves before everyone else: the 5 Distributed Inference repos climbing fastest, every Tuesday.

Free weekly digest. Double opt-in, unsubscribe anytime.

Maintain twinspark?

Claim this page to add your own call-to-action.

Claim this page