# LLM Torrents

> LLM Torrents is a catalog of open-weight large language models distributed over BitTorrent, with hardware fit guidance, quantization info, and setup help. Downloads are free and require no account.

## Key pages
- [Models](https://llmtorrents.com/models): browse and filter the model catalog
- [Collections](https://llmtorrents.com/collections): editorial lists (e.g. best under 8 GB, best coding models)
- [Learn](https://llmtorrents.com/learn): explainers on quant formats, VRAM requirements, and context/KV-cache
- [Calculator](https://llmtorrents.com/calculator): check whether a model fits your hardware
- [Break-even calculator](https://llmtorrents.com/calculator/break-even): local GPU vs. cloud rental — when buying pays off
- [Budget GPU picker](https://llmtorrents.com/calculator/budget): what to buy for a given budget
- [Quant helper](https://llmtorrents.com/quants/helper): pick a GPU and model, rank its quants by quality and fit
- [Wizard](https://llmtorrents.com/wizard): guided model recommendation
- [AMD & Intel GPU viability status board](https://llmtorrents.com/gpu/amd-intel-status): dated, sourced status of ROCm / Vulkan / SYCL-IPEX for local LLMs on AMD & Intel
- [Alternatives](https://llmtorrents.com/alternatives): closest local open-weight models to Claude, GPT and Gemini + the hardware to run them
- [Changelog](https://llmtorrents.com/changelog): recent model drops and site updates
- [Open data](https://llmtorrents.com/open-data): nightly downloadable catalog dump (JSON + checksum, CC0)
- [Seed with the llmt CLI](https://llmtorrents.com/seeding/guide#install-cli): command-line tool that scans models you already have, verifies their SHA-256 against the catalog, and seeds them without re-downloading
- [Trust & safety](https://llmtorrents.com/trust): what we mirror, what "verified" means here, how takedowns work, and privacy

## Collections
- [Best under 8 GB](https://llmtorrents.com/collections/best-under-8gb)
- [Best coding (≤ 24 GB)](https://llmtorrents.com/collections/best-coding-24gb)
- [Best on Apple Silicon](https://llmtorrents.com/collections/best-on-apple-silicon)

## Explainers
- [Quantization formats explained](https://llmtorrents.com/learn/quantization-formats)
- [VRAM guide: what fits in 8, 12, 16 and 24 GB](https://llmtorrents.com/learn/vram-guide)
- [Context and the KV cache](https://llmtorrents.com/learn/context-and-kv-cache)
- [What does "abliterated" mean?](https://llmtorrents.com/learn/abliterated-uncensored-models)
- [Bandwidth vs compute: why tokens/sec tracks memory bandwidth](https://llmtorrents.com/learn/bandwidth-vs-compute)
- [MoE vs dense: why a 30B-A3B runs near 8B speed](https://llmtorrents.com/learn/moe-vs-dense)
- [Prompt processing vs token generation](https://llmtorrents.com/learn/prompt-processing-vs-token-generation)
- [Which quant should I pick?](https://llmtorrents.com/learn/which-quant-should-i-pick)
- [K-quants, I-quants and imatrix](https://llmtorrents.com/learn/k-quants-vs-i-quants-imatrix)
- [NVIDIA, AMD and Intel: the local-LLM software stacks](https://llmtorrents.com/learn/gpu-software-stacks)
- [Which Mac chip for local LLMs?](https://llmtorrents.com/learn/which-mac-chip-for-llms)
- [Ollama vs llama.cpp vs vLLM vs LM Studio — which and when](https://llmtorrents.com/learn/local-llm-runtimes)
- [Multi-GPU for local LLMs: when a second card is worth it](https://llmtorrents.com/learn/multi-gpu-local-llms)
- [Project ideas by hardware tier: what to actually run locally](https://llmtorrents.com/learn/project-ideas-by-hardware-tier)
- [The smallest usable local LLM setup: 8 GB RAM, CPU only](https://llmtorrents.com/learn/smallest-usable-local-llm-setup)

## Local alternatives to frontier models
- [Local alternatives to Claude Opus](https://llmtorrents.com/alternatives/claude-opus)
- [Local alternatives to Claude Sonnet](https://llmtorrents.com/alternatives/claude-sonnet)
- [Local alternatives to GPT (latest)](https://llmtorrents.com/alternatives/gpt-latest)
- [Local alternatives to Gemini (latest)](https://llmtorrents.com/alternatives/gemini-latest)

## Answers
- [Answers hub](https://llmtorrents.com/answers): what runs on your hardware, by VRAM budget, Mac, or use case
- [What runs on 6 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/6gb-vram)
- [What runs on 8 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/8gb-vram)
- [What runs on 12 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/12gb-vram)
- [What runs on 16 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/16gb-vram)
- [What runs on 24 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/24gb-vram)
- [What runs on 32 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/32gb-vram)
- [What runs on 48 GB of VRAM](https://www.llmtorrents.com/what-can-i-run/48gb-vram)
- [What runs on a 8 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/8gb)
- [What runs on a 16 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/16gb)
- [What runs on a 24 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/24gb)
- [What runs on a 36 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/36gb)
- [What runs on a 48 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/48gb)
- [What runs on a 64 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/64gb)
- [What runs on a 128 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/128gb)
- [What runs on a 192 GB Mac](https://www.llmtorrents.com/what-can-i-run/mac/192gb)
- [Coding models that fit 16 GB](https://www.llmtorrents.com/models-for/coding/16gb)
- [Coding models that fit 24 GB](https://www.llmtorrents.com/models-for/coding/24gb)
- [Coding models that fit 32 GB](https://www.llmtorrents.com/models-for/coding/32gb)
- [Coding models that fit 48 GB](https://www.llmtorrents.com/models-for/coding/48gb)
- [RAG models that fit 6 GB](https://www.llmtorrents.com/models-for/rag/6gb)
- [RAG models that fit 8 GB](https://www.llmtorrents.com/models-for/rag/8gb)
- [RAG models that fit 12 GB](https://www.llmtorrents.com/models-for/rag/12gb)
- [RAG models that fit 16 GB](https://www.llmtorrents.com/models-for/rag/16gb)
- [RAG models that fit 24 GB](https://www.llmtorrents.com/models-for/rag/24gb)
- [RAG models that fit 32 GB](https://www.llmtorrents.com/models-for/rag/32gb)
- [RAG models that fit 48 GB](https://www.llmtorrents.com/models-for/rag/48gb)