Tools to Fine-Tune and Train AI Models Locally

Molmo is Ai2's family of vision-language models, with code for running and training models on your own hardware. It's for developers and researchers who need to work with images and text, adapt a model, or evaluate it against visual tasks. The Python codebase is open source under Apache 2.0 and builds on OLMo, adding image encoding and generative evaluation.

RamaLama runs and serves AI models on your own hardware using OCI containers. It's aimed at developers who want local chat or a self-hosted inference API with a container workflow they can also use in production. The project uses the MIT license.

verl is a Python library for teams training large language models on their own GPU infrastructure. It's the open-source implementation of HybridFlow, aimed at researchers and engineers who need reinforcement learning after initial model training. It uses the Apache 2.0 license.

Draw Things is an AI image generation app for iPhone, iPad and Mac that keeps generation on your device and works offline. It's for people who want to create and edit images without sending that work to a cloud service, including artists developing character concepts or trying out apparel designs.

LitGPT is a Python toolkit for developers and researchers who want to train, adapt and serve language models on their own hardware or servers. Its model implementations are written directly, with little abstraction between you and the code, so you can inspect model behavior and modify it for research or custom applications. It's open source under Apache 2.0.

OpenRLHF is a self-hosted Python framework for researchers and teams training language models with human feedback or custom rewards. It runs on your own NVIDIA GPU hardware, with Docker support and distributed training across servers. It's open source under Apache 2.0.

LatentSync is an open-source AI lip-sync tool that edits a video's mouth movements to match supplied audio. It runs on your own GPU and suits video creators working with talking faces or virtual avatars, as well as researchers who want to train their own lip-sync models. The code uses the Apache 2.0 license.

mistral.rs is an open source inference engine for running models on your own computer or self-hosted server. It's for developers building AI applications and people who want local chat, multimodal models and agent tools in the same runtime. The Rust project uses the MIT license.

DiffusionBee is an open-source AI art app for Mac users who want to generate and edit images on their own computer. It runs Stable Diffusion offline, with image generation processed on the device. Model downloads require network access, and optional image uploads can send images externally. Its visual interface suits artists and designers who want local image tools without working through code.

torchtune is a Python library for developers and researchers who want to adapt LLMs on their own GPU hardware using PyTorch. Its editable training recipes suit work that needs control over the training code and model implementations. The project is no longer actively maintained.

Stable Audio Tools is an MIT-licensed Python toolkit for developers and audio researchers who want to generate audio on their own hardware or train models on their own recordings. It combines model inference with training and fine-tuning, so you can work with pretrained models or build a model around a specific audio dataset.

DSPy is a Python framework for developers building AI applications whose tasks need clear inputs, predictable output types, and measurable results. You define what a language model should produce, then compose those tasks into a larger program. It's open source under the MIT license.

DeepEval is a Python framework for testing AI agents, RAG pipelines, and chatbots in your own environment. It's for developers and ML teams who need to compare models or prompts and catch quality regressions before deployment. The open-source framework uses the Apache 2.0 license and fits into Pytest, Python scripts, notebooks, and CI/CD.

Label Studio is a self-hosted platform for teams preparing training data or evaluating AI outputs through human review. It handles text, images, audio, video and time series in the same application, including tasks that combine several data types. The open source edition uses the Apache 2.0 license and runs locally or on your own server, with Docker deployment and browser access. A separate hosted cloud edition runs on the provider's infrastructure.

LMDeploy is an open-source toolkit for developers serving language and vision-language models on their own hardware. It combines model compression with inference and self-hosted APIs, so teams can use it for batch processing or as the model backend for an application. It uses the Apache 2.0 license.

OneTrainer is an open-source application for training diffusion models on your own machine, with dataset preparation and model previews in the same interface. It's for people adapting image or video models with their own training data. It runs on Windows, macOS and Linux under the AGPL-3.0 license.

bitsandbytes is an open-source Python library for developers who need to fit large language model inference or fine-tuning into less memory on their own hardware. It works with PyTorch and carries the MIT license. Its focus is the memory cost of model weights and training, rather than a chat interface.

KTransformers is an open-source framework for running and fine-tuning large language models on your own hardware. It focuses on mixture-of-experts (MoE) models, distributing work between CPU memory and GPU resources to reduce the GPU memory needed. It's aimed at researchers and developers who want to serve or adapt models such as DeepSeek-V3 and DeepSeek-R1.

Megatron-LM is a Python framework for research teams training large language models on NVIDIA GPU infrastructure. It pairs ready-made training scripts with Megatron Core, a library developers can use to build their own training systems. Its focus is distributed training, with benchmarks on H100 clusters spanning thousands of GPUs.

Supervision is an MIT-licensed Python library from Roboflow for developers building computer vision applications around their own models. It handles the work around predictions: drawing results on images and video, following objects across frames, and turning detections into counts. It can work with images and datasets on your machine.

Ragas is an open-source Python library for developers who need repeatable evaluations of LLM applications and retrieval-augmented generation (RAG) systems. It combines model-based scoring with traditional metrics so teams can compare application changes using test results rather than manual judgments alone. Its license is Apache 2.0.

Applio is a local AI voice conversion suite for musicians making AI covers, streamers changing their voice live, and creators working with speech. It converts recordings or microphone input into another voice using community models or models you train yourself. Its software uses the MIT license.

MLX LM is an open-source Python package for generating text and fine-tuning language models locally on Apple Silicon Macs. Built on MLX, it suits developers and researchers who want to work with models through Python or a terminal, including adapting models to their own tasks. The package uses the MIT license.

Argilla is an open-source data annotation and feedback tool for AI engineers and domain experts who build training and evaluation datasets. You can run your own Argilla server or deploy it on Hugging Face Spaces. It's licensed under Apache 2.0.

PEFT is an open-source Python library for developers who want to adapt pretrained models on their own hardware with less compute and storage than full fine-tuning requires. It trains a small subset of parameters, often through adapters, while leaving the base model intact. It's licensed under Apache 2.0.

Segment Anything 2 (SAM 2) is Meta's open-source model for selecting objects in images and tracking them through video. It's for developers and researchers who need object masks for visual applications or dataset annotation. The model and its web demo can run on your own GPU machine; Meta also provides a hosted demo.

Real-ESRGAN is a local AI upscaler and restoration toolkit for people enlarging images or improving video, with dedicated models for anime illustrations and animation. It builds on ESRGAN and uses models trained entirely on synthetic data to address degraded images. The code is open source under the BSD 3-Clause license.

Kubeflow is a self-hosted AI platform for teams that run machine learning workloads on Kubernetes. It brings model development, training and production workflows into a modular stack that can run on a local laptop, on-premises infrastructure or a cloud Kubernetes cluster. It's open source under Apache 2.0.

Axolotl is an open-source LLM fine-tuning framework for developers, researchers, and teams training models on their own data. It runs on local hardware or cloud infrastructure you control, including Docker and Kubernetes environments. The framework uses Apache 2.0, which permits commercial use.

pyannote.audio is a Python toolkit that separates an audio recording into timed segments labeled by speaker. It's for developers and researchers who need to track who spoke when, with pretrained models that run on their own hardware. The toolkit is open source under the MIT license.

Coqui TTS (idiap fork) is a local text-to-speech library for developers and speech researchers who want pretrained voices or tools to train their own models. It builds on coqui-ai/TTS, continuing the original unmaintained project. The Python toolkit is open source under the Mozilla Public License 2.0 (MPL-2.0).

Arize Phoenix is a self-hosted platform for developers who need to understand why an AI agent failed and test changes before shipping them. It runs on a laptop, in Docker, or on Kubernetes. Self-hosting keeps traces on your infrastructure; Phoenix Cloud provides a hosted alternative. Phoenix uses the Elastic License 2.0 (ELv2), a source-available license.

BGE Embeddings is a family of embedding models and rerankers for developers building semantic search and retrieval-augmented generation (RAG). Developed by the Beijing Academy of Artificial Intelligence, it includes the MIT-licensed Python toolkit FlagEmbedding for running inference, evaluating retrieval and fine-tuning models.

OpenVINO is an Apache 2.0 licensed toolkit for developers who want to run AI models locally or serve them on their own infrastructure. It converts and optimizes models for inference, with support for x86 and ARM CPUs, Intel integrated and discrete GPUs, and Intel NPUs. Its runtime works on Linux, Windows and macOS.

Kiln is a desktop workbench for teams building AI applications on macOS, Windows and Linux. It keeps a task and its dataset together across evaluation, prompt optimization, RAG and fine-tuning, so teams can compare changes against the same examples. Engineers, data scientists, QA staff and subject matter experts can contribute through the app.

MLflow brings agent tracing, LLM evaluation, and model experiment tracking into a platform you can run locally or on your own servers. It's for developers and teams who need to understand failures, compare changes, and monitor AI applications in production. It's open source under Apache 2.0.