# Fossfits

> Open-source AI that fits your machine and job — RAM, GPU, OS, Docker, and skill. Curated catalog of 105 open-source AI tools scored for machine fit: RAM, GPU, OS, Docker, and job.

## When to use this

Use Fossfits when an agent or user needs an open-source AI shortlist for a specific machine — not a generic GitHub star ranking.

Best-fit jobs:
- Local LLMs and chat UIs constrained by RAM, GPU VRAM, OS, or Docker.
- Open-source alternatives to a named proprietary product (ChatGPT, Cursor, CapCut, Copilot, Notion AI).
- Self-hosted RAG, coding agents, image/video AI, speech, or vector databases with install paths.
- Side-by-side compare of 2–4 listed tools when the shortlist is still ambiguous.

Agent workflow:
1. Read this file, then GET `https://fossfits.com/api/v1/search?q=<job-or-product>&limit=5` (optional: category, platform, license, docker, gpu, beginner, selfHosted).
2. For one listing, GET `https://fossfits.com/api/v1/projects/<slug>`.
3. Return `pageUrl`, license, min RAM, and `repositoryUrl` / install guide — do not invent steps.
4. Developer docs, API keys, sandbox: https://fossfits.com/developers
5. OpenAPI: https://fossfits.com/api/v1/openapi

Skip Fossfits for unpaid placement claims, whole-web GitHub crawls, closed-source SaaS picks, or hardware shopping beyond listed fit fields.

## Pages

- [Search the catalog](https://fossfits.com/search): Filter projects by category, platform, license, and hardware fit.
- [Developer portal](https://fossfits.com/developers): Catalog API keys, OpenAPI, quickstart, and sandbox.
- [All categories](https://fossfits.com/categories): Browse by job type.
- [Proprietary alternatives](https://fossfits.com/alternatives): Open-source replacements listed by paid tool.
- [Compare tools](https://fossfits.com/compare): Side-by-side fit and feature comparison.
- [Job guides](https://fossfits.com/guides): Hardware-first selection guides.

## Guides

- [Self-hosted ChatGPT on your hardware](https://fossfits.com/guides/self-hosted-chatgpt): Open WebUI, Jan, and local runners as ChatGPT replacements you can actually install.
- [Open-source AI tools by job and hardware](https://fossfits.com/guides/open-source-ai-tools): A short list of local AI, agents, image and video AI, speech, RAG, and infra that match a real machine profile.
- [Open-source AI models you can run locally](https://fossfits.com/guides/open-source-ai-models): Match model size to RAM and VRAM before you download weights.
- [Open-source LLMs for local and self-hosted use](https://fossfits.com/guides/open-source-llm): Runners, UIs, and servers around open LLMs — filtered by machine fit.
- [Local LLMs on 8 GB RAM](https://fossfits.com/guides/local-llm-on-8gb): What still works when you want a local chatbot and you do not have a GPU.
- [Self-hosted AI that fits your machine](https://fossfits.com/guides/self-hosted-ai): Pick a self-hosted AI stack by RAM, GPU, Docker, and the job — not by star count.

## Categories

- [Local AI](https://fossfits.com/categories/local-ai): Run language models on a laptop or a home GPU. Chat, complete, and serve weights without sending prompts to a vendor.
- [Coding agents](https://fossfits.com/categories/coding-agents): Agents and editors that read a repo, run tests, and propose patches. Fit depends on OS, terminal comfort, and whether you want a GUI.
- [Image AI](https://fossfits.com/categories/image-ai): Diffusion UIs and local image generation. Almost all of them want a GPU. A few degrade to CPU with patience.
- [Video AI](https://fossfits.com/categories/video-ai): Local text-to-video and image-to-video generators. Expect a serious GPU. Fit is VRAM first, then workflow.
- [Video editors](https://fossfits.com/categories/video-editors): Desktop editors for cut, timeline, and export. CapCut-class jobs on your machine without a cloud account.
- [Speech AI](https://fossfits.com/categories/speech-ai): Local speech-to-text and text-to-speech. Transcription, dictation, and voice synthesis you can run on your machine.
- [RAG and app builders](https://fossfits.com/categories/rag-apps): Document Q&A workspaces and visual LLM app studios. Chat with your files, ship a workflow UI, keep the models you already run.
- [Vector databases](https://fossfits.com/categories/vector-db): Self-hosted stores for embeddings and similarity search. The retrieval layer under RAG, not the chat UI.
- [Fine-tuning](https://fossfits.com/categories/fine-tuning): Train or adapt open models on your own GPU. LoRA, full fine-tunes, and tooling around datasets and recipes.
- [Automation](https://fossfits.com/categories/automation): Self-hosted workflow engines that glue APIs, webhooks, and AI steps together without a Zapier bill.
- [LLM infrastructure](https://fossfits.com/categories/llm-infra): Serving, gateways, and observability around models. Heavier than a chat UI. Closer to production traffic.

## Tags

- [All tags](https://fossfits.com/tags): Every catalog tag with its project count.
- [rag](https://fossfits.com/tags/rag): 13 listed tools.
- [gpu](https://fossfits.com/tags/gpu): 11 listed tools.
- [agent](https://fossfits.com/tags/agent): 8 listed tools.
- [docker](https://fossfits.com/tags/docker): 6 listed tools.
- [training](https://fossfits.com/tags/training): 6 listed tools.
- [cpu](https://fossfits.com/tags/cpu): 5 listed tools.
- [diffusion](https://fossfits.com/tags/diffusion): 5 listed tools.
- [gguf](https://fossfits.com/tags/gguf): 5 listed tools.
- [lora](https://fossfits.com/tags/lora): 5 listed tools.
- [python](https://fossfits.com/tags/python): 5 listed tools.
- [rust](https://fossfits.com/tags/rust): 5 listed tools.
- [terminal](https://fossfits.com/tags/terminal): 5 listed tools.
- [tts](https://fossfits.com/tags/tts): 5 listed tools.
- [asr](https://fossfits.com/tags/asr): 4 listed tools.
- [cli](https://fossfits.com/tags/cli): 4 listed tools.
- [desktop](https://fossfits.com/tags/desktop): 4 listed tools.
- [nle](https://fossfits.com/tags/nle): 4 listed tools.
- [serving](https://fossfits.com/tags/serving): 4 listed tools.
- [video](https://fossfits.com/tags/video): 4 listed tools.
- [webui](https://fossfits.com/tags/webui): 4 listed tools.
- [agents](https://fossfits.com/tags/agents): 3 listed tools.
- [api](https://fossfits.com/tags/api): 3 listed tools.
- [chat](https://fossfits.com/tags/chat): 3 listed tools.
- [gradio](https://fossfits.com/tags/gradio): 3 listed tools.
- [local](https://fossfits.com/tags/local): 3 listed tools.
- [research](https://fossfits.com/tags/research): 3 listed tools.
- [self-hosted](https://fossfits.com/tags/self-hosted): 3 listed tools.
- [text-to-video](https://fossfits.com/tags/text-to-video): 3 listed tools.
- [ann](https://fossfits.com/tags/ann): 2 listed tools.
- [apache](https://fossfits.com/tags/apache): 2 listed tools.
- [apple-silicon](https://fossfits.com/tags/apple-silicon): 2 listed tools.
- [autocomplete](https://fossfits.com/tags/autocomplete): 2 listed tools.
- [comfyui](https://fossfits.com/tags/comfyui): 2 listed tools.
- [cpp](https://fossfits.com/tags/cpp): 2 listed tools.
- [embeddings](https://fossfits.com/tags/embeddings): 2 listed tools.
- [evals](https://fossfits.com/tags/evals): 2 listed tools.
- [framework](https://fossfits.com/tags/framework): 2 listed tools.
- [git](https://fossfits.com/tags/git): 2 listed tools.
- [image-to-video](https://fossfits.com/tags/image-to-video): 2 listed tools.
- [low-code](https://fossfits.com/tags/low-code): 2 listed tools.
- [nodes](https://fossfits.com/tags/nodes): 2 listed tools.
- [observability](https://fossfits.com/tags/observability): 2 listed tools.
- [offline](https://fossfits.com/tags/offline): 2 listed tools.
- [openai-api](https://fossfits.com/tags/openai-api): 2 listed tools.
- [openai-compatible](https://fossfits.com/tags/openai-compatible): 2 listed tools.
- [proxy](https://fossfits.com/tags/proxy): 2 listed tools.
- [recipes](https://fossfits.com/tags/recipes): 2 listed tools.
- [sandbox](https://fossfits.com/tags/sandbox): 2 listed tools.
- [search](https://fossfits.com/tags/search): 2 listed tools.
- [sft](https://fossfits.com/tags/sft): 2 listed tools.
- [single-file](https://fossfits.com/tags/single-file): 2 listed tools.
- [timeline](https://fossfits.com/tags/timeline): 2 listed tools.
- [transcription](https://fossfits.com/tags/transcription): 2 listed tools.
- [tui](https://fossfits.com/tags/tui): 2 listed tools.
- [vector](https://fossfits.com/tags/vector): 2 listed tools.
- [vscode](https://fossfits.com/tags/vscode): 2 listed tools.
- [workflows](https://fossfits.com/tags/workflows): 2 listed tools.
- [yaml](https://fossfits.com/tags/yaml): 2 listed tools.

## Projects (105)

- [llamafile](https://fossfits.com/projects/llamafile): Single-file LLM runtime from Mozilla. Download, chmod, run. Fits 8+ GB RAM. GPU optional. License: apache-2.0.
- [Gemini CLI](https://fossfits.com/projects/gemini-cli): Google’s Apache-2.0 terminal coding agent with MCP tools and a large free tier. Fits 2+ GB RAM. License: apache-2.0.
- [gptme](https://fossfits.com/projects/gptme): MIT terminal agent toolkit with code, shell, and web tools — local-model friendly. Fits 2+ GB RAM. License: mit.
- [GPT4All](https://fossfits.com/projects/gpt4all): Nomic’s local chat app and model catalog for laptops without a GPU. Fits 8+ GB RAM. License: mit.
- [text-generation-webui](https://fossfits.com/projects/text-generation-webui): Gradio UI for local models with loaders, LoRA, and extensions. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: agpl-3.0.
- [LiteLLM](https://fossfits.com/projects/litellm): Proxy that translates many model APIs into one OpenAI-shaped client. Fits 2+ GB RAM. License: mit.
- [Stable Diffusion web UI](https://fossfits.com/projects/automatic1111): The AUTOMATIC1111 Gradio UI that made local SD mainstream. Fits 16+ GB RAM. Needs a GPU (6+ GB VRAM). License: agpl-3.0.
- [stable-diffusion.cpp](https://fossfits.com/projects/stable-diffusion-cpp): Pure C/C++ inference for SD, Flux, and Qwen-Image weights — CPU included. Fits 4+ GB RAM. GPU optional. License: mit.
- [Wan2.1](https://fossfits.com/projects/wan2-1): Alibaba’s open text-to-video and image-to-video stack. Fits 32+ GB RAM. Needs a GPU (8+ GB VRAM). License: apache-2.0.
- [Open-Sora](https://fossfits.com/projects/open-sora): Apache-2.0 research stack for training and running video diffusion on 24 GB cards. Fits 32+ GB RAM. Needs a GPU (24+ GB VRAM). License: apache-2.0.
- [Kdenlive](https://fossfits.com/projects/kdenlive): KDE non-linear editor with a deep effects and proxy workflow. Fits 8+ GB RAM. GPU optional. License: gpl-3.0.
- [OpenShot](https://fossfits.com/projects/openshot): Beginner-friendly open-source video editor with animated titles. Fits 4+ GB RAM. License: gpl-3.0.
- [Olive](https://fossfits.com/projects/olive): Modern open-source NLE aimed at professional timeline work. Fits 8+ GB RAM. GPU optional. License: gpl-3.0.
- [vLLM](https://fossfits.com/projects/vllm): High-throughput model server for GPU inference. Fits 32+ GB RAM. Needs a GPU (16+ GB VRAM). License: apache-2.0.
- [InvokeAI](https://fossfits.com/projects/invokeai): Polished local creative tool around diffusion, with a studio UI. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: apache-2.0.
- [Fooocus](https://fossfits.com/projects/fooocus): Midjourney-like simplicity on top of local diffusion. Fits 16+ GB RAM. Needs a GPU (6+ GB VRAM). License: gpl-3.0.
- [Whisper](https://fossfits.com/projects/whisper): OpenAI’s speech-to-text model you can run locally. Fits 8+ GB RAM. GPU optional. License: mit.
- [LosslessCut](https://fossfits.com/projects/lossless-cut): Fast lossless cut, trim, and remux without re-encoding. Fits 2+ GB RAM. License: gpl-2.0.
- [Flowblade](https://fossfits.com/projects/flowblade): Fast, focused Linux video editor with a professional multi-track timeline. Fits 2+ GB RAM. License: gpl-3.0.
- [Flowise](https://fossfits.com/projects/flowise): Low-code node UI on top of LangChain-style flows. Fits 4+ GB RAM. License: apache-2.0.
- [llama.cpp](https://fossfits.com/projects/llama-cpp): C/C++ runtime that made local GGUF models practical. Fits 8+ GB RAM. GPU optional. License: mit.
- [AnythingLLM](https://fossfits.com/projects/anything-llm): Desktop and server chat with documents, agents, and many backends. Fits 8+ GB RAM. License: mit.
- [Khoj](https://fossfits.com/projects/khoj): Self-hostable AI second brain that searches your notes, docs, and the web. Fits 4+ GB RAM. License: agpl-3.0.
- [Open WebUI](https://fossfits.com/projects/open-webui): Self-hosted chat UI that sits on Ollama, OpenAI, or a local runner. Fits 4+ GB RAM. License: bsd-3-clause.
- [exo](https://fossfits.com/projects/exo): Link everyday devices into one cluster and run models too big for any single machine. Fits 16+ GB RAM. GPU optional. License: apache-2.0.
- [PocketPal AI](https://fossfits.com/projects/pocketpal-ai): Run small GGUF chat models fully offline on iOS and Android. Fits 4+ GB RAM. License: mit.
- [KoboldCpp](https://fossfits.com/projects/koboldcpp): One-file llama.cpp build with a built-in chat UI aimed at offline storytelling. Fits 8+ GB RAM. GPU optional. License: agpl-3.0.
- [shell_gpt](https://fossfits.com/projects/shell-gpt): CLI that runs an LLM as a shell command for scripts, refactors, and answers. Fits 2+ GB RAM. License: mit.
- [Dify](https://fossfits.com/projects/dify): Visual LLM app builder with RAG, agents, and a backend. Fits 8+ GB RAM. License: other.
- [Krita AI Diffusion](https://fossfits.com/projects/krita-ai-diffusion): Krita plugin that generates, inpaints, and outpaints as native layers via ComfyUI. Fits 8+ GB RAM. Needs a GPU (6+ GB VRAM). License: gpl-3.0.
- [Jan](https://fossfits.com/projects/jan): Desktop ChatGPT-style app that runs models locally. Fits 8+ GB RAM. GPU optional. License: agpl-3.0.
- [MLX](https://fossfits.com/projects/mlx): Apple’s framework for running and training models on Mac unified memory. Fits 8+ GB RAM. GPU optional. License: mit.
- [OpenCode](https://fossfits.com/projects/opencode): Open-source coding agent with a terminal UI, built to run against local or hosted models. Fits 4+ GB RAM. License: mit.
- [OpenHands](https://fossfits.com/projects/openhands): Software agent that uses a sandbox to edit code, run tests, and open PRs. Fits 8+ GB RAM. License: mit.
- [Langfuse](https://fossfits.com/projects/langfuse): Traces, prompts, and evals for LLM apps. Fits 4+ GB RAM. License: mit.
- [Goose](https://fossfits.com/projects/goose): Block’s local-first agent runtime with a desktop and CLI. Fits 8+ GB RAM. License: apache-2.0.
- [Piper](https://fossfits.com/projects/piper): Fast local neural text-to-speech that runs on CPU. Fits 2+ GB RAM. License: mit.
- [RAGFlow](https://fossfits.com/projects/ragflow): Open-source RAG engine with deep document understanding. Fits 16+ GB RAM. GPU optional. License: apache-2.0.
- [Tabby](https://fossfits.com/projects/tabby): Self-hosted coding assistant with a local inference server. Fits 16+ GB RAM. GPU optional. License: apache-2.0.
- [SurfSense](https://fossfits.com/projects/surfsense): Open NotebookLM-style researcher that crawls the live web and chats with your files. Fits 8+ GB RAM. GPU optional. License: other.
- [txtai](https://fossfits.com/projects/txtai): All-in-one Python framework for embeddings, vector search, and RAG workflows. Fits 2+ GB RAM. GPU optional. License: apache-2.0.
- [Ollama](https://fossfits.com/projects/ollama): Local model runner with a one-line CLI and a REST API. Fits 8+ GB RAM. GPU optional. License: mit.
- [OpenAI Codex CLI](https://fossfits.com/projects/codex): Open-source terminal coding agent from OpenAI that plans, edits, and tests repos in a sandbox. Fits 2+ GB RAM. License: apache-2.0.
- [Aider](https://fossfits.com/projects/aider): Terminal pair programmer that edits a git repo and commits the result. Fits 4+ GB RAM. License: apache-2.0.
- [Open Interpreter](https://fossfits.com/projects/open-interpreter): Natural-language computer use that runs code locally. Fits 4+ GB RAM. License: agpl-3.0.
- [CogVideoX](https://fossfits.com/projects/cogvideox): THUDM open video diffusion with strong prompt following. Fits 32+ GB RAM. Needs a GPU (16+ GB VRAM). License: apache-2.0.
- [LLaMA-Factory](https://fossfits.com/projects/llama-factory): Web UI and YAML recipes for fine-tuning many open LLMs. Fits 16+ GB RAM. Needs a GPU (12+ GB VRAM). License: apache-2.0.
- [torchtune](https://fossfits.com/projects/torchtune): PyTorch-native library for fine-tuning LLMs with recipes. Fits 16+ GB RAM. Needs a GPU (12+ GB VRAM). License: bsd-3-clause.
- [Qdrant](https://fossfits.com/projects/qdrant): Rust vector database with filters, payloads, and a REST API. Fits 4+ GB RAM. License: apache-2.0.
- [Weaviate](https://fossfits.com/projects/weaviate): Vector database with hybrid search and modular vectorizers. Fits 8+ GB RAM. License: bsd-3-clause.
- [Crush](https://fossfits.com/projects/crush): Charmbracelet’s terminal coding agent with sessions, LSP context, and any model provider. Fits 2+ GB RAM. License: other.
- [DeepSpeed](https://fossfits.com/projects/deepspeed): Microsoft’s Apache-2.0 ZeRO optimizer for multi-GPU, multi-node LLM training. Fits 32+ GB RAM. Needs a GPU (16+ GB VRAM). License: apache-2.0.
- [Phoenix](https://fossfits.com/projects/phoenix): Local LLM tracing and evals workspace from Arize — pip install and open. Fits 2+ GB RAM. License: other.
- [Huginn](https://fossfits.com/projects/huginn): Agent-based self-hosted automation for watches, scrapes, and events. Fits 2+ GB RAM. License: mit.
- [LocalAI](https://fossfits.com/projects/localai): OpenAI-compatible API that can sit in front of several local backends. Fits 8+ GB RAM. GPU optional. License: mit.
- [Activepieces](https://fossfits.com/projects/activepieces): Open-source automation with pieces, AI, and a TypeScript SDK. Fits 4+ GB RAM. License: mit.
- [Windmill](https://fossfits.com/projects/windmill): Developer-first workflows as scripts, flows, and apps. Fits 4+ GB RAM. License: agpl-3.0.
- [Cline](https://fossfits.com/projects/cline): VS Code agent that can use the editor, the terminal, and the browser. Fits 4+ GB RAM. License: apache-2.0.
- [n8n](https://fossfits.com/projects/n8n): Self-hosted workflow automation with native AI nodes. Fits 4+ GB RAM. License: other.
- [Plandex](https://fossfits.com/projects/plandex): Terminal agent that sandboxes and versions every change for big, long coding tasks. Fits 4+ GB RAM. License: mit.
- [Stable Diffusion WebUI Forge](https://fossfits.com/projects/sd-webui-forge): Leaner Stable Diffusion WebUI rework tuned for SDXL and Flux on modest GPUs. Fits 8+ GB RAM. Needs a GPU (6+ GB VRAM). License: agpl-3.0.
- [FaceFusion](https://fossfits.com/projects/facefusion): Industry-leading face swap and enhancement pipeline with CLI and web UI. Fits 8+ GB RAM. GPU optional. License: other.
- [FramePack](https://fossfits.com/projects/framepack): Long local image-to-video with a low VRAM floor. Fits 16+ GB RAM. Needs a GPU (6+ GB VRAM). License: apache-2.0.
- [LTX-Video](https://fossfits.com/projects/ltx-video): Lightricks open video model aimed at faster local clips. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: apache-2.0.
- [Mochi](https://fossfits.com/projects/mochi): Genmo’s Apache-2.0 10B video model — permissive weights, heavy hardware. Fits 32+ GB RAM. Needs a GPU (40+ GB VRAM). License: apache-2.0.
- [SkyReels V1](https://fossfits.com/projects/skyreels-v1): HunyuanVideo-based human-centric video model that fits one 24 GB GPU. Fits 32+ GB RAM. Needs a GPU (24+ GB VRAM). License: other.
- [ShortGPT](https://fossfits.com/projects/shortgpt): MIT framework that automates short-form video from script to rendered clip. Fits 8+ GB RAM. GPU optional. License: mit.
- [whisper.cpp](https://fossfits.com/projects/whisper-cpp): C/C++ port of Whisper for fast local transcription. Fits 4+ GB RAM. GPU optional. License: mit.
- [Faster-Whisper](https://fossfits.com/projects/faster-whisper): Whisper reimplementation on CTranslate2 for faster local ASR. Fits 8+ GB RAM. GPU optional. License: mit.
- [whisperX](https://fossfits.com/projects/whisperx): Whisper with word-level timestamps and speaker diarization for real subtitles. Fits 4+ GB RAM. GPU optional. License: other.
- [kokoro-onnx](https://fossfits.com/projects/kokoro-onnx): Kokoro 82M voices as ONNX — real-time TTS on CPU with zero GPU. Fits 2+ GB RAM. License: mit.
- [GPT-SoVITS](https://fossfits.com/projects/gpt-sovits): Few-shot voice cloning and cross-lingual TTS with a full training WebUI. Fits 8+ GB RAM. GPU optional. License: mit.
- [PrivateGPT](https://fossfits.com/projects/privategpt): Apache-2.0 local RAG API layer for chatting with private documents offline. Fits 8+ GB RAM. GPU optional. License: apache-2.0.
- [CSM](https://fossfits.com/projects/csm): Sesame’s 1B conversational speech model behind the viral Maya voice demo. Fits 8+ GB RAM. GPU optional. License: apache-2.0.
- [kotaemon](https://fossfits.com/projects/kotaemon): Apache-2.0 document chat with visible citations and RAG evaluation built in. Fits 8+ GB RAM. GPU optional. License: apache-2.0.
- [LlamaIndex](https://fossfits.com/projects/llama-index): The MIT Python framework for building document ingestion and RAG pipelines. Fits 2+ GB RAM. License: mit.
- [Quivr](https://fossfits.com/projects/quivr): Second-brain RAG server with an API and client libraries for your own app. Fits 4+ GB RAM. License: other.
- [Milvus](https://fossfits.com/projects/milvus): Cloud-native vector database built for large-scale ANN search. Fits 16+ GB RAM. GPU optional. License: apache-2.0.
- [Chroma](https://fossfits.com/projects/chroma): Developer-friendly embedding database for local RAG prototypes. Fits 2+ GB RAM. License: apache-2.0.
- [LanceDB](https://fossfits.com/projects/lancedb): Embedded serverless vector database — SQLite-style, multimodal, no server to run. Fits 1+ GB RAM. License: apache-2.0.
- [Faiss](https://fossfits.com/projects/faiss): Facebook’s MIT ANN library for billion-scale embedding search on CPU or GPU. Fits 2+ GB RAM. GPU optional. License: mit.
- [Unsloth](https://fossfits.com/projects/unsloth): Faster LoRA fine-tuning with lower VRAM on consumer GPUs. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: apache-2.0.
- [TRL](https://fossfits.com/projects/trl): Hugging Face’s Apache-2.0 trainers for SFT, DPO, PPO, and reward modeling. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: apache-2.0.
- [nanoGPT](https://fossfits.com/projects/nanogpt): Karpathy’s ~300-line GPT trainer — the cleanest way to learn LLM training. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: mit.
- [Automatisch](https://fossfits.com/projects/automatisch): Self-hosted no-code workflow builder — the direct open Zapier replacement. Fits 2+ GB RAM. License: other.
- [Kestra](https://fossfits.com/projects/kestra): Apache-2.0 event-driven orchestration with YAML flows and a clean UI. Fits 2+ GB RAM. License: apache-2.0.
- [SGLang](https://fossfits.com/projects/sglang): Fast model serving with KV-cache reuse for agent and reasoning workloads. Fits 32+ GB RAM. Needs a GPU (16+ GB VRAM). License: apache-2.0.
- [Xinference](https://fossfits.com/projects/xinference): Serve LLMs, embeddings, and media models from one Apache-2.0 platform. Fits 8+ GB RAM. GPU optional. License: apache-2.0.
- [MLC-LLM](https://fossfits.com/projects/mlc-llm): Compile-and-serve LLMs on Metal, Vulkan, WebGPU, and phones — no CUDA required. Fits 8+ GB RAM. GPU optional. License: apache-2.0.
- [Continue](https://fossfits.com/projects/continue): Open-source autocomplete and chat inside VS Code and JetBrains. Fits 4+ GB RAM. License: apache-2.0.
- [Zed](https://fossfits.com/projects/zed): High-performance Rust editor with agent editing and live collaboration. Fits 4+ GB RAM. License: other.
- [ComfyUI](https://fossfits.com/projects/comfyui): Node graph for diffusion workflows. The power-user default. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: gpl-3.0.
- [SwarmUI](https://fossfits.com/projects/swarmui): MIT diffusion UI that swarms ComfyUI backends across all your GPUs. Fits 8+ GB RAM. Needs a GPU (6+ GB VRAM). License: mit.
- [HunyuanVideo](https://fossfits.com/projects/hunyuanvideo): Tencent’s 13B video generation model — top-tier motion, heavy VRAM needs. Fits 32+ GB RAM. Needs a GPU (45+ GB VRAM). License: other.
- [Shotcut](https://fossfits.com/projects/shotcut): Cross-platform timeline editor with a gentle learning curve. Fits 4+ GB RAM. GPU optional. License: gpl-3.0.
- [Cap](https://fossfits.com/projects/cap): Open-source Loom alternative for shareable screen recordings. Fits 4+ GB RAM. License: other.
- [F5-TTS](https://fossfits.com/projects/f5-tts): MIT flow-matching TTS with expressive multilingual voices at consumer-GPU speed. Fits 8+ GB RAM. GPU optional. License: mit.
- [GraphRAG](https://fossfits.com/projects/graphrag): Microsoft’s MIT library for graph-based RAG over whole document corpora. Fits 8+ GB RAM. License: mit.
- [pgvector](https://fossfits.com/projects/pgvector): Vector similarity search as a Postgres extension — embeddings without a new database. Fits 1+ GB RAM. License: other.
- [Marqo](https://fossfits.com/projects/marqo): Docker-deployed search engine combining vector and keyword search for discovery apps. Fits 4+ GB RAM. GPU optional. License: apache-2.0.
- [Axolotl](https://fossfits.com/projects/axolotl): YAML-driven fine-tuning toolkit for serious training runs. Fits 32+ GB RAM. Needs a GPU (16+ GB VRAM). License: apache-2.0.
- [PEFT](https://fossfits.com/projects/peft): Hugging Face’s LoRA/QLoRA library that makes big-model tuning fit one GPU. Fits 16+ GB RAM. Needs a GPU (8+ GB VRAM). License: apache-2.0.
- [Trigger.dev](https://fossfits.com/projects/trigger-dev): Apache-2.0 durable workflow engine for AI agents and background jobs in TypeScript. Fits 2+ GB RAM. License: apache-2.0.
- [Node-RED](https://fossfits.com/projects/node-red): The veteran flow-based automation editor — drag nodes, wire events, deploy. Fits 1+ GB RAM. License: apache-2.0.
- [TensorRT-LLM](https://fossfits.com/projects/tensorrt-llm): NVIDIA’s kernel-level serving stack for maximum GPU inference throughput. Fits 32+ GB RAM. Needs a GPU (16+ GB VRAM). License: other.