whichllm

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.

demo

See it

$ whichllm --gpu "RTX 4090"

#1  Qwen/Qwen3.6-27B     27.8B  Q5_K_M   score 92.8    27 t/s
#2  Qwen/Qwen3-32B       32.0B  Q4_K_M   score 83.0    31 t/s
#3  Qwen/Qwen3-30B-A3B   30.0B  Q5_K_M   score 82.7   102 t/s

The 32B model fits your card fine — whichllm still ranks the 27B #1, because it scores higher on real benchmarks and is a newer generation. A size-only "what fits?" tool would hand you the bigger one. That gap is the whole point of whichllm. (Note #3: a MoE model at 102 t/s — speed is ranked on active params, quality on total.)

What can I run?

Real top picks (snapshot 2026-05 — your results track live HuggingFace data, this is not a static list):

Hardware	VRAM	Top pick	Speed
RTX 5090	32 GB	`Qwen3.6-27B` · Q6_K · score 94.7	~40 t/s
RTX 4090 / 3090	24 GB	`Qwen3.6-27B` · Q5_K_M · score 92.8	~27 t/s
RTX 4060	8 GB	`Qwen3-14B` · Q3_K_M · score 71.0	~22 t/s
Apple M3 Max	36 GB	`Qwen3.6-27B` · Q5_K_M · score 89.4	~9 t/s
CPU only	—	`gpt-oss-20b` (MoE) · Q4_K_M · score 45.2	~6 t/s

whichllm --gpu "<your card>" to simulate any of these before you buy.

Useful? A GitHub star helps other people find it — and I'd genuinely like to know what it picked for your rig: drop it in Issues.

Star History

Why whichllm?

Fitting a model into your VRAM is the easy part. The hard part is knowing which of the models that fit is actually the best — and that is what whichllm is built to get right.

Evidence-based ranking, not a size heuristic — The top pick is chosen from merged real benchmarks (LiveBench, Artificial Analysis, Aider, multimodal/vision, Chatbot Arena ELO, Open LLM Leaderboard) — never "the biggest model that happens to fit."
Recency-aware — Stale leaderboards are demoted along each model's lineage, so a 2024 model can't outrank a current-generation one on an outdated score. The benchmark snapshot date is printed under every ranking, so a stale recommendation is self-evident instead of silently trusted.
Evidence-graded and guarded — Every score is tagged direct / variant / base / interpolated / self-reported and discounted by confidence. Fabricated uploader claims and cross-family inheritance (a small fork borrowing its much larger base's score) are actively rejected.
Architecture-aware estimates — VRAM = weights + GQA KV cache + activation + overhead; speed is bandwidth-bound with per-quant efficiency, per-backend factors, MoE active-vs-total split, and unified-memory vs discrete-PCIe partial-offload modeling.
One command, scriptable — whichllm prints the answer; add --json | jq for pipelines. No TUI, no keybindings to memorize.
Live data — Models fetched directly from the HuggingFace API, with curated frozen fallbacks for offline or rate-limited use.

Features

Auto-detect hardware — NVIDIA, AMD, Apple Silicon, CPU-only
Smart ranking — Scores models by VRAM fit, speed, and benchmark quality
One-command chat — whichllm run downloads and starts a chat session instantly
Code snippets — whichllm snippet prints ready-to-run Python for any model
Live data — Fetches models directly from HuggingFace (cached for performance)
Benchmark-aware — Integrates real eval scores with confidence-based dampening
Task profiles — Filter by general, coding, vision, or math use cases
GPU simulation — Test with any GPU: whichllm --gpu "RTX 4090"
Hardware planning — Reverse lookup: whichllm plan "llama 3 70b"
JSON output — Pipe-friendly: whichllm --json

Run & Snippet

Try any model with a single command. No manual installs needed — whichllm creates an isolated environment via uv, installs dependencies, downloads the model, and starts an interactive chat.

run demo

# Chat with a model (auto-picks the best GGUF variant)
whichllm run "qwen 2.5 1.5b gguf"

# Auto-pick the best model for your hardware and chat
whichllm run

# CPU-only mode
whichllm run "phi 3 mini gguf" --cpu-only

Works with all model formats:

GGUF — via llama-cpp-python (lightweight, fast)
AWQ / GPTQ — via transformers + autoawq / auto-gptq
FP16 / BF16 — via transformers

Get a copy-paste Python snippet instead:

whichllm snippet "qwen 7b"

from llama_cpp import Llama

llm = Llama.from_pretrained(
    repo_id="Qwen/Qwen2.5-7B-Instruct-GGUF",
    filename="qwen2.5-7b-instruct-q4_k_m.gguf",
    n_ctx=4096,
    n_gpu_layers=-1,
    verbose=False,
)

output = llm.create_chat_completion(
    messages=[{"role": "user", "content": "Hello!"}],
)
print(output["choices"][0]["message"]["content"])

Install

uv (recommended)

uvx whichllm

To install permanently:

uv tool install whichllm

Homebrew

brew install andyyyy64/whichllm/whichllm

pip

pip install whichllm

Development

git clone https://github.com/Andyyyy64/whichllm.git
cd whichllm
uv sync --dev
uv run whichllm
uv run pytest

Usage

# Auto-detect hardware and show best models
whichllm

# Simulate a GPU (e.g. planning a purchase)
whichllm --gpu "RTX 4090"
whichllm --gpu "RTX 5090"
# Specify variant
whichllm --gpu "RTX 5060 16"


# CPU-only mode
whichllm --cpu-only

# More results / filters
whichllm --top 20
whichllm --quant Q4_K_M
whichllm --min-speed 30
whichllm --evidence base   # allow id/base-model matches
whichllm --evidence strict # id-exact only (same as --direct)
whichllm --direct

# JSON output
whichllm --json

# Force refresh (ignore cache)
whichllm --refresh

# Show hardware info only
whichllm hardware

# Plan: what GPU do I need for a specific model?
whichllm plan "llama 3 70b"
whichllm plan "Qwen2.5-72B" --quant Q8_0
whichllm plan "mistral 7b" --context-length 32768

# Run: download and chat with a model instantly
whichllm run "qwen 2.5 1.5b gguf"
whichllm run                       # auto-pick best for your hardware

# Snippet: print ready-to-run Python code
whichllm snippet "qwen 7b"
whichllm snippet "llama 3 8b gguf" --quant Q5_K_M

Integrations

Ollama

Find the best model and run it directly:

# Pick the top model and run it with Ollama
whichllm --top 1 --json | jq -r '.models[0].model_id' | xargs ollama run

# Find the best coding model
whichllm --profile coding --top 1 --json | jq -r '.models[0].model_id' | xargs ollama run

Shell alias

Add to your .bashrc / .zshrc:

alias bestllm='whichllm --top 1 --json | jq -r ".models[0].model_id"'
# Usage: ollama run $(bestllm)

Scoring

Each model gets a 0-100 score. Benchmark quality and size form the core; evidence confidence and runtime fit then scale it, with speed, source trust, and popularity as adjustments.

Factor	Effect	Description
Benchmark quality	core	Merged LiveBench / Artificial Analysis / Aider / Vision / Arena ELO / Open LLM Leaderboard, weighted by source confidence
Model size	up to 35	`log2`-scaled world-knowledge proxy (MoE uses total params)
Quantization	× penalty	Lower-bit quants discounted multiplicatively
Evidence confidence	×0.55–1.0	none / self-reported ×0.55, inherited ×0.78, direct full
Runtime fit	×0.50–1.0	partial-offload ×0.72, CPU-only ×0.50
Speed	-8 to +8	Usability gate vs a fit-dependent tok/s floor
Source trust	-5 to +5	Official-org bonus, known-repackager penalty
Popularity	tie-breaker	Downloads/likes; weight shrinks as evidence strengthens

Score markers:

~ (yellow) — No direct benchmark; score inherited/interpolated from the model family
? (yellow) — No benchmark data available

How it works

Data pipeline

Model fetching — Fetches popular models from HuggingFace API:
- Text-generation (downloads + recently updated)
- GGUF-filtered (separate query for coverage)
- Vision models (image-text-to-text) when --profile vision or any
Benchmark sources — Current tier (LiveBench, Artificial Analysis Index, Aider) merged live when reachable, plus a curated multimodal / vision index; frozen tier (Open LLM Leaderboard v2, Chatbot Arena ELO). Tiers have separate caps and lineage-aware recency demotion so stale leaderboards stop over-rewarding older generations.
Benchmark evidence — Five resolution levels, increasingly discounted:
- direct — Exact model ID match
- variant — Suffix-stripped or -Instruct variant
- base_model — Base model from cardData
- line_interp — Size-aware interpolation within model family
- self_reported — Uploader-claimed eval (heavily discounted)
Inheritance is rejected when a model's params diverge more than 2× from its family's dominant member, catching draft / MTP / abliterated forks that share a family_id with a much larger base.
Cache — ~/.cache/whichllm/:
- models.json — 6h TTL
- benchmark.json — 24h TTL

Ranking engine

Hardware detection — NVIDIA (nvidia-ml-py), AMD (dbgpu/ROCm), Apple Silicon (Metal), CPU cores, RAM, disk
VRAM estimation — Weights + KV cache + activation + framework overhead (~500MB)
Compatibility — Full GPU / Partial Offload / CPU-only; compute capability and OS checks
Speed — tok/s from GPU memory bandwidth lookup (constants.py)
Scoring — Benchmark (with confidence dampening), size, quantization penalty, fit type, speed, popularity, source trust (official vs repackager)
Backend filter — Apple Silicon and CPU-only restrict to GGUF for stability; Linux+NVIDIA allows AWQ/GPTQ

Project structure

src/whichllm/
├── cli.py              # Typer CLI: main, plan, run, snippet, hardware
├── constants.py        # GPU bandwidth, quantization bytes, compute capability
├── hardware/
│   ├── detector.py     # Orchestrates GPU/CPU/RAM detection
│   ├── nvidia.py       # NVIDIA GPU via nvidia-ml-py
│   ├── amd.py          # AMD GPU (Linux)
│   ├── apple.py        # Apple Silicon (Metal)
│   ├── cpu.py          # CPU name, cores, AVX support
│   ├── memory.py       # RAM and disk free
│   ├── gpu_simulator.py # --gpu flag: synthetic GPU from name
│   └── types.py        # GPUInfo, HardwareInfo
├── models/
│   ├── fetcher.py      # HuggingFace API, model parsing, evalResults
│   ├── benchmark.py    # Arena ELO, Leaderboard (parquet/rows API)
│   ├── grouper.py      # Family grouping by base_model and name
│   ├── cache.py        # JSON cache with TTL
│   └── types.py        # ModelInfo, GGUFVariant, ModelFamily
├── engine/
│   ├── vram.py         # VRAM = weights + KV cache + activation + overhead
│   ├── compatibility.py# Fit type, disk check, compute/OS warnings
│   ├── performance.py  # tok/s from bandwidth
│   ├── quantization.py # Bytes per weight, quality penalty, non-GGUF inference
│   ├── ranker.py       # Scoring, evidence filter, profile/match
│   └── types.py        # CompatibilityResult
└── output/
    └── display.py      # Rich table, JSON output, hardware/plan displays

Contributing

Contributions are welcome! See CONTRIBUTING.md for guidelines.

Requirements

Python 3.11+
NVIDIA GPU detection via nvidia-ml-py (included by default)
AMD / Apple Silicon detected automatically

License

MIT