BYOAIK (Bring Your Own AI Key) is a directory of AI tools that run on your own API key.

torchtune

PyTorch-native fine-tuning recipes with no framework layer between you and the training loop.

Website Source code

Category:
Fine-Tuning and Training
Maintenance:
Slowing down (last commit about 5 months ago)
Pricing:
Open Source
Open source:
Yes (BSD-3-Clause)
Self-hostable:
Yes
Local-first:
Yes
Platforms:
CLI, Self-hosted, Local-only, Docker
AI providers (bring your own key):
Ollama, Custom / OpenAI-compatible
API key storage:
No cloud key required for local model
Key risk level:
LOW
Trust score:
91/100

torchtune is Meta's PyTorch-native library of fine-tuning recipes, written to be read and modified rather than configured, which suits people who want to see exactly what the training loop does. It runs locally with no key and no service. Note the pace: the last commit on the default branch was several months before review, so it is steady rather than busy, and the maintenance badge on this listing reflects that. BSD-3-Clause.

Why this trust score (91/100)

Trust measures how the tool treats your API key and how much of that has been verified. It contains no popularity signal.

  • Key Safety 25/25: No cloud key is needed at all. The model runs locally.
  • Request Routing 17/20: Requests go straight from you to the AI provider.
  • Transparency 20/20: Source is public under BSD-3-Clause, so anyone can check how the key is handled. Key handling was located in the published source.
  • Privacy 14/15: Local-first: it works without sending your data anywhere. Can be self-hosted, so the data path stays inside infrastructure you control.
  • Maintenance 8/10: Commits within the last six months.
  • Verification Confidence 7/10: Key handling was found in the published source by BYOAIK's scanner.

What was checked

Verification tier SOURCE_VERIFIED, derived from the evidence below and not set by hand.

  • [REPORTED · RESEARCH] This listing was compiled by an AI-assisted research pass over the tool's public website, README and documentation. No person independently confirmed it.
  • [STRONG · SOURCE_SCAN] The user supplies their own Hugging Face key: the project references its API key variable. source
  • [STRONG · SOURCE_SCAN] Supports a local model backend, so it can run with no cloud provider key at all. source
  • [CONFIRMED · SOURCE_SCAN] Most recent commit 2026-04-23, about 5 month(s) ago. source

How your API key is handled

Key handling varies. Requests are sent directly to the AI provider. Because it can be self-hosted, your key never has to touch a third-party backend.

Setup

Add your provider API key in settings and choose a model.