# CanIRunLocal

> Free checker that tells you which open-weight LLMs (Llama, Qwen, Gemma, Mistral, DeepSeek, gpt-oss and more) fit on a given GPU or Apple Silicon Mac, at which quantization, roughly how fast, with copy-paste setup for Ollama, LM Studio, llama.cpp, MLX or vLLM. No account needed. Made by OnDevice Labs (https://x.com/OnDeviceLabs). Data updated 2026-10-14.

## How to query it
- Interactive results for a machine: https://canirunlocal.si/?hw=<hardware name>, e.g. https://canirunlocal.si/?hw=RTX+3060+12GB or https://canirunlocal.si/?hw=M4+Pro+48GB (the name must match one of the hardware names listed below; spaces as + or %20).
- Pick a runtime with &rt=ollama|lmstudio|llamacpp|mlx|vllm, e.g. https://canirunlocal.si/?hw=M4+Max+128GB&rt=mlx
- The interactive page needs JavaScript. For plain-HTML answers, use the static page per machine: https://canirunlocal.si/hw/<slug> (list below), or the full text dump at https://canirunlocal.si/llms-full.txt.
- Raw data: https://canirunlocal.si/models.json (Ollama models, sizes per tag) and https://canirunlocal.si/runtimes.json (Hugging Face builds for other runtimes).

## How "fits" is decided
Model download size (from ollama.com) plus headroom for ~8k tokens of context, compared with usable fast memory: VRAM on a PC GPU, or about 67-75% of unified memory on a Mac. The highest-quality quantization that fits is picked. On PC GPUs a model may "partially" fit by offloading to system RAM, which is much slower. Speeds are estimates from memory bandwidth, not benchmarks.

## Hardware pages
- [RTX 3060 12GB](https://canirunlocal.si/hw/rtx-3060-12gb)
- [RTX 3060 Ti 8GB](https://canirunlocal.si/hw/rtx-3060-ti-8gb)
- [RTX 3070 8GB](https://canirunlocal.si/hw/rtx-3070-8gb)
- [RTX 3080 10GB](https://canirunlocal.si/hw/rtx-3080-10gb)
- [RTX 3090 24GB](https://canirunlocal.si/hw/rtx-3090-24gb)
- [RTX 4060 8GB](https://canirunlocal.si/hw/rtx-4060-8gb)
- [RTX 4060 Ti 16GB](https://canirunlocal.si/hw/rtx-4060-ti-16gb)
- [RTX 4070 12GB](https://canirunlocal.si/hw/rtx-4070-12gb)
- [RTX 4070 Ti Super 16GB](https://canirunlocal.si/hw/rtx-4070-ti-super-16gb)
- [RTX 4080 16GB](https://canirunlocal.si/hw/rtx-4080-16gb)
- [RTX 4090 24GB](https://canirunlocal.si/hw/rtx-4090-24gb)
- [RTX 5070 12GB](https://canirunlocal.si/hw/rtx-5070-12gb)
- [RTX 5080 16GB](https://canirunlocal.si/hw/rtx-5080-16gb)
- [RTX 5090 32GB](https://canirunlocal.si/hw/rtx-5090-32gb)
- [RX 7600 8GB](https://canirunlocal.si/hw/rx-7600-8gb)
- [RX 7800 XT 16GB](https://canirunlocal.si/hw/rx-7800-xt-16gb)
- [RX 7900 XTX 24GB](https://canirunlocal.si/hw/rx-7900-xtx-24gb)
- [RX 9070 XT 16GB](https://canirunlocal.si/hw/rx-9070-xt-16gb)
- [M1 8GB](https://canirunlocal.si/hw/m1-8gb)
- [M1 16GB](https://canirunlocal.si/hw/m1-16gb)
- [M1 Pro 32GB](https://canirunlocal.si/hw/m1-pro-32gb)
- [M1 Max 64GB](https://canirunlocal.si/hw/m1-max-64gb)
- [M2 8GB](https://canirunlocal.si/hw/m2-8gb)
- [M2 16GB](https://canirunlocal.si/hw/m2-16gb)
- [M2 24GB](https://canirunlocal.si/hw/m2-24gb)
- [M2 Pro 32GB](https://canirunlocal.si/hw/m2-pro-32gb)
- [M2 Max 64GB](https://canirunlocal.si/hw/m2-max-64gb)
- [M2 Ultra 192GB](https://canirunlocal.si/hw/m2-ultra-192gb)
- [M3 8GB](https://canirunlocal.si/hw/m3-8gb)
- [M3 16GB](https://canirunlocal.si/hw/m3-16gb)
- [M3 24GB](https://canirunlocal.si/hw/m3-24gb)
- [M3 Pro 36GB](https://canirunlocal.si/hw/m3-pro-36gb)
- [M3 Max 64GB](https://canirunlocal.si/hw/m3-max-64gb)
- [M3 Max 128GB](https://canirunlocal.si/hw/m3-max-128gb)
- [M4 16GB](https://canirunlocal.si/hw/m4-16gb)
- [M4 24GB](https://canirunlocal.si/hw/m4-24gb)
- [M4 32GB](https://canirunlocal.si/hw/m4-32gb)
- [M4 Pro 48GB](https://canirunlocal.si/hw/m4-pro-48gb)
- [M4 Max 64GB](https://canirunlocal.si/hw/m4-max-64gb)
- [M4 Max 128GB](https://canirunlocal.si/hw/m4-max-128gb)
- [Mac Studio M4 Max (32c GPU) 36GB](https://canirunlocal.si/hw/mac-studio-m4-max-32c-gpu-36gb)
- [Mac Studio M4 Max (40c GPU) 48GB](https://canirunlocal.si/hw/mac-studio-m4-max-40c-gpu-48gb)
- [Mac Studio M4 Max (40c GPU) 64GB](https://canirunlocal.si/hw/mac-studio-m4-max-40c-gpu-64gb)
- [Mac Studio M4 Max (40c GPU) 128GB](https://canirunlocal.si/hw/mac-studio-m4-max-40c-gpu-128gb)
- [Mac Studio M3 Ultra 96GB](https://canirunlocal.si/hw/mac-studio-m3-ultra-96gb)
- [Mac Studio M3 Ultra 256GB](https://canirunlocal.si/hw/mac-studio-m3-ultra-256gb)
- [Mac Studio M3 Ultra (80c GPU) 512GB](https://canirunlocal.si/hw/mac-studio-m3-ultra-80c-gpu-512gb)
- [Mac Studio M5 Max (32c GPU) 36GB](https://canirunlocal.si/hw/mac-studio-m5-max-32c-gpu-36gb)
- [Mac Studio M5 Max (40c GPU) 48GB](https://canirunlocal.si/hw/mac-studio-m5-max-40c-gpu-48gb)
- [Mac Studio M5 Max (40c GPU) 64GB](https://canirunlocal.si/hw/mac-studio-m5-max-40c-gpu-64gb)
- [Mac Studio M5 Max (40c GPU) 128GB](https://canirunlocal.si/hw/mac-studio-m5-max-40c-gpu-128gb)
- [Mac Studio M5 Ultra 96GB](https://canirunlocal.si/hw/mac-studio-m5-ultra-96gb)
- [Mac Studio M5 Ultra (64c GPU) 256GB](https://canirunlocal.si/hw/mac-studio-m5-ultra-64c-gpu-256gb)
- [Mac Studio M5 Ultra (80c GPU) 256GB](https://canirunlocal.si/hw/mac-studio-m5-ultra-80c-gpu-256gb)
- [DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte)](https://canirunlocal.si/hw/dgx-spark-128gb-nvidia-asus-dell-lenovo-hp-msi-acer-gigabyte)
- [DGX Spark 64GB (ASUS, Dell, HP, MSI, Acer, Gigabyte)](https://canirunlocal.si/hw/dgx-spark-64gb-asus-dell-hp-msi-acer-gigabyte)

## Optional
- [Full per-hardware recommendations as text](https://canirunlocal.si/llms-full.txt)