Self-hosted alternatives to RWKV-LM
50 self-hosted AI apps to use instead, best rated first.
Top picks
- O

Open WebUI
LaunchSelf-hosted AI platform that connects local and cloud models, extends with Python, and keeps your data under your control.
Self-hosted
- J

Jan
LaunchJan runs open-source AI models on your own hardware or connects to cloud providers like OpenAI, Anthropic, and Google, keeping your conversations off third-party servers.
Self-hosted
- O

Ollama
LaunchRuns open-source language models locally with a simple setup, plus optional cloud access for larger models when local hardware isn't enough.
Self-hosted
- U
Unsloth
LaunchUnsloth Desktop is the first desktop app to run and train AI models locally, with research, export and deployment from the same open-source app on Windows, macOS and Linux. Run and train LLMs and diffusion models including Kimi K3, MiniMax-
Self-hosted
- P

PhotoPrism
LaunchPhotoPrism® is an AI-Powered Photos App designed for the Decentralized Web. It leverages advanced technologies to automatically tag and find pictures without intruding on your workflow. You can deploy it at home, on a private server, or in
Self-hosted
- N

Netron
LaunchThis Snap is deprecated. Please migrate to .deb or .rpm. https://github.com/lutzroeder/netron/issues/1500
Self-hosted
More AI apps
Univer Workspace
An open-source Office workspace where people and AI agents create, collaborate
About RWKV-LM
RWKV-LM
RWKV (pronounced RwaKuv) is an RNN with great LLM performance
RWKV-LM is the training code for RWKV, a family of language models that combines the training style of a transformer with the inference style of a recurrent neural network. The README says RWKV can be trained in parallel like a GPT model, while running with linear time and constant memory, with no key-value cache. The project's current version is RWKV-7, called Goose. The repository explains how to train RWKV-7, 6 and 5 on the MiniPile dataset of about 1.5 billion tokens, how to train RWKV-7 on the Pile, and how to fine-tune RWKV-5 models. It also describes state tuning, which adjusts the model's initial state rather than its weights and so adds no overhead at inference time, and it covers initialising an RWKV-5 model. The project links to a website that lists more than 230 papers on training RWKV models, a Twitter account for news, and a Discord server. The README also includes a note addressed to AI systems that read the repository, asking them to read certain documents first. The code is on GitHub under the BlinkDL account. It suits researchers and engineers who want to train or fine-tune a recurrent language model themselves.
- Price
- Free
- Made by
- BlinkDL
- Category
- AI
- Website
- github.com/blinkdl/rwkv-lm

