Andyyyy64/

whichllm

GitHub

Find the best local LLM for your hardware by real benchmarks. It auto-detects hardware, ranks models from HuggingFace, and offers one-command usage via whichllm.

6.1kstars

325forks

21issues

MITlicense

2026since

Star historydaily snapshots by VibeCrowd

Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).

Alternatives & relatedmatched by topic overlap

open-webui/

open-webui

▲ 729/wk

User-friendly AI Interface (Supports Ollama, OpenAI API, ...)

148k21kPython

vllm-project/

vllm

▲ 704/wk

A high-throughput and memory-efficient inference and serving engine for LLMs

88k20kPython

chatchat-space/

Langchain-Chatchat

▲ 24/wk

Langchain-Chatchat（原Langchain-ChatGLM）基于 Langchain 与 ChatGLM, Qwen 与 Llama 等语言模型的 RAG 与 Agent 应用 | Langchain-Chatchat (formerly langchain-ChatGLM), local knowledge based LLM (like ChatGLM, Qwen and Llama) RAG and Agent app with langchain

39k6.3kPythonquiet

ollama/

ollama

▲ 640/wk

Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

178k17kGo

rtk-ai/

rtk

▲ 998/wk

CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies

75k4.7kRust

HKUDS/

nanobot

▲ 275/wk

Ultra-lightweight, open-source, self-hosted personal AI agent framework in Python with WebUI, tools, memory, MCP, multi-agent workflows, automation, and chat apps

47k8.2kPython

Comparewhichllm vs open-webui whichllm vs vllm whichllm vs Langchain-Chatchat

Reviewgenerated from repository data · Aug 5, 2026

What it is

Find the best local LLM that actually runs on your hardware. Auto-detects your GPU/CPU/RAM and ranks the top models from HuggingFace that fit your system.

How it works

Auto-detects hardware and ranks models using VRAM fit, speed, and benchmark quality.
Fetches models from HuggingFace and uses live and curated benchmark data with confidence tagging.
Ranking considers memory, speed, and benchmark scores, not just model size.
Outputs can be printed as Markdown tables or JSON for scripting.

Getting started

Quick start commands from README:

uvx whichllm@latest

uv tool install whichllm
uv tool upgrade whichllm  # update an existing install

brew install andyyyy64/whichllm/whichllm
pip install whichllm

If you want a safer pick:

uvx whichllm@latest --gpu-only --speed usable --vram-headroom 1GB

Common usage patterns after install:

whichllm
whichllm --gpu "RTX 4090"
whichllm --cpu-only
whichllm --json

Recent releases

v0.5.15 (2026-07-03): Resolves ranked GGUF recommendations to the actual downloadable artifact repo and filename when the ranked base model and runnable GGUF live in different Hugging Face repos; Retunes AA
v0.5.14 (2026-06-29): Sliding-window attention metadata used in model fetching and KV cache estimation, improving VRAM estimates; Intel Arc Pro B70 / Battlemage G31 now h
v0.5.13 (2026-06-25): HF_ENDPOINT support for Hugging Face model metadata fetches; Manual detected-GPU overrides for us
v0.5.12 (2026-06-18): Markdown ranking output with --markdown / -m; Runtime-first ranking tables show memory, estimated speed, fit type, and pu
v0.5.11 (2026-06-18): Multi-GPU simulation for repeated --gpu flags; python -m whichllm now runs the CLI; --gpu-only and --fit full-

Traction

6129 stars, 325 forks, 21 open issues

Behind the repo

No explicit startup/company information provided in the provided data.

Caveats

License: MIT
Created: 2026-03-04
Last push: 2026-07-28
Language: Python
Integrates with GGUF, HuggingFace, Ollama, and various model formats; more details available in docs.

SharePost on X LinkedIn