Self-hosted alternatives to Sunk Cost
200 self-hosted apps like Sunk Cost, closest match first.
Top picks
- L

LocalAI
LaunchLocalAI is an open-source, MIT-licensed AI engine that runs language models, speech, vision, image and video generation and agents on your own hardware. Its API is compatible with OpenAI, and the site says it also speaks the Anthropic, Olla
FreeOpen sourceSelf-hosted
- T

TurboLLM
LaunchRun open AI models on your own machine — no cloud, no account, fully offline. Any GGUF model on any llama.cpp-compatible engine, auto-tuned to your hardware, with real measured tokens/sec before you download, a fast browser UI, and an OpenA
FreeOpen sourceSelf-hosted
- G

Guaardvark
LaunchSelf-hosted AI studio on your own GPU: chat with your files (RAG), screen and browser agents, coding swarms, LoRA training, an MCP server for Claude Code and Cursor, and local video, image, music and voice generation. No telemetry.
FreeOpen sourceSelf-hosted
- L
Llmaker
LaunchModels, vector DBs, embeddings, a retrieval agent, and evals: the whole modern LLM stack on your own hardware, from one command. OpenAI-compatible API + web UI per instance, GPU partitioning, no data leaving your machine.
FreeOpen sourceSelf-hosted
- S
Self-Hosted AI Stack
LaunchDeploy a self-hosted AI stack with Docker Compose: local LLMs, chat, document parsing, embeddings, transcription, text-to-speech and MCP tools. Powered by Ollama, LiteLLM, AnythingLLM and more. Local-first with auto-configured authenticatio
FreeOpen sourceSelf-hosted
- N
NadirClaw
LaunchSource-available LLM router and AI cost optimizer. Routes simple prompts to cheap or local models and complex ones to premium models, automatically. Drop-in OpenAI-compatible proxy for Claude Code, Codex, Cursor and OpenClaw. Self-hosted, n
FreeOpen sourceSelf-hosted
More like Sunk Cost
Self Host N8n On GCR
Self-host n8n on Google Cloud without the subscription fees or server headaches
About Sunk Cost
Sunk Cost
How long until local AI pays for itself?
Sunk Cost is a calculator that works out how long a machine for running local AI models takes to pay for itself, compared with paying for an API. You pick a machine such as a Mac mini, Mac Studio, DGX Spark or Strix Halo box, set how many tokens you use a day, the context window and the input-to-output ratio, and the site shows which open models fit and how fast they run. The break-even view can be adjusted for electricity cost, API speed, your own generation speed or your monthly bill. Where nothing has been measured, local speed is estimated from memory bandwidth and labelled as an estimate. The site is candid about the results. At 500k tokens a day on the strongest model, the quickest pay-back in its table is 26 years, on a GeForce RTX 3060 with 12GB, and nothing in the table pays back inside a decade at that usage. The table covers 56 machines, and the site also has best-buy lists by usage level. Your inputs are sent once you stop typing, with no cookies and no IP address.
- Category
- AI
- Website
- sunkcost.ai



