Sign in

Compare repositories

This page shows the Star History and the facts of 2 to 5 repositories in the Repository Index, side by side. You choose the repositories. The page does not rank them.

Stars gained: From Aug 15, 2026 to Nov 17, 2026, each from zeroStars gainedFrom Aug 15, 2026 to Nov 17, 2026, each from zero01K2KAug 15Nov 17, 2026vllm-project/speculatorsraullenchai/Rapid-MLXsyv-ai/HyperQwenyoussofal/MTPLXSafeAILab/EAGLE
Stars: The star count at the end of each dayStarsThe star count at the end of each day02K4KMay 8Nov 17, 2026vllm-project/speculatorsraullenchai/Rapid-MLXsyv-ai/HyperQwenyoussofal/MTPLXSafeAILab/EAGLE

Facts

Fact vllm-project/speculators

A unified library for building, evaluating, and storing speculative decoding algorithms for LLM inference in vLLM

raullenchai/Rapid-MLX

Rapid-MLX is an open-source (Apache 2.0) OpenAI- and Anthropic-compatible LLM inference server for Apple Silicon, built on MLX, focused on reliable tool calling for coding agents. Up to 4× faster than Apple's MLX (mlx-lm) on the same weights.

syv-ai/HyperQwen

Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.

youssofal/MTPLX

The fastest way to run Qwen 3.8 Flash Next, Qwen 3.8 27B and Ternary Bonsai 2 27B on a Mac: 125 tok/s in OpenCode on an M5 Max, and a 27B model on 16 GB Macs. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anthropic compatible local server.

SafeAILab/EAGLE

Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3 (NeurIPS'25).

Stars 9614,1582,7432,7202,579
Net stars, last 30 days +74+171+669+129+26
Largest star day +22 on Jul 2, 2026+79 on May 8, 2026+128 on Aug 25, 2026+87 on May 16, 2026+9 on Aug 2, 2026
Milestone reached in the record 500 on Jun 11, 2026None1,000 on Sep 1, 20261,000 on Jul 13, 2026None
Forks 295445341213298
Language PythonPythonPythonPythonPython
License APACHE-2.0APACHE-2.0APACHE-2.0APACHE-2.0Other license
Created Apr 8, 2025Feb 25, 2026Aug 15, 2026May 2, 2026Dec 7, 2023
Last push Nov 17, 2026Nov 18, 2026Nov 10, 2026Nov 7, 2026Feb 20, 2026
Archived NoNoNoNoNo
Health facts: Maintenance 2 of 4 checks (2 without data)2 of 4 checks (2 without data)3 of 4 checks2 of 4 checks (2 without data)1 of 4 checks (2 without data)
Health facts: Releases 2 of 3 checks (1 without data)2 of 3 checks (1 without data)0 of 3 checks (1 without data)2 of 3 checks (1 without data)0 of 3 checks (2 without data)
Health facts: Security practice 0 of 1 check (1 without data)0 of 1 check (1 without data)0 of 1 check0 of 1 check (1 without data)0 of 1 check (1 without data)
Health facts: Community files 1 of 3 checks (2 without data)1 of 3 checks (2 without data)1 of 3 checks1 of 3 checks (2 without data)1 of 3 checks (2 without data)
Commits, last 13 weeks No dataNo data749No dataNo data
Weeks with commits, last 52 No dataNo data13 of 52No dataNo data
In the Repository Index since Nov 15, 2026Nov 10, 2026Oct 23, 2026Nov 11, 2026Nov 11, 2026

Topics that all of them have on the Repository Host: speculative-decoding. The facts come from the Repository Host and from the Star History that AwesomeRepos records. A shared topic does not mean that one repository can replace another. Each Health Facts row is one dimension alone. AwesomeRepos does not add them into one score. How each check is defined

Example: oven-sh/bun, denoland/deno, nodejs/node