Rwkv Runner
A RWKV management and startup tool, full automation, only 8MB. And provides an interface compatible with the OpenAI API. RWKV is a large language model that is fully open source and available for commercial use.
Last verified:
What is Rwkv Runner?
Rwkv Runner is a lightweight desktop GUI application for managing and launching RWKV large language models with full automation. It is a management and startup tool that eliminates barriers to using LLMs by providing a single executable program of only 8MB. The tool includes an OpenAI API-compatible interface, meaning any ChatGPT client can work as an RWKV client, and it comes with built-in chat, completion, and composition interaction interfaces.
Key features include RWKV model management with one-click startup, automatic dependency installation, built-in WebUI for one-click web service start, built-in model conversion tool, built-in download management and remote model inspection, and one-click LoRA finetuning (Windows Only). It supports front-end and back-end separation deployment, preset multi-level VRAM configs, WebGPU strategy for AMD/Intel GPUs, MIDI hardware input with track editing, chat presets, attachment uploads, multilingual localization, theme switching, and automatic updates. The tool can also function as a client for OpenAI ChatGPT, GPT-Playground, and Ollama.
Rwkv Runner is designed for developers, researchers, and anyone wanting to use RWKV language models locally without complex setup. It works on almost all computers with pre-set VRAM configurations, supports CUDA acceleration for NVIDIA GPUs, and allows backend deployment on a server with the program acting as a client. The tool is fully open source under MIT license and supports RWKV models including the latest RWKV-7
Rwkv Runner pricing
Pricing model: Freemium
RWKV-Runner is completely free and open source under MIT license. RWKV is fully open source and available for commercial use. No paid plans, subscription tiers, or fees are mentioned. The tool is a lightweight executable requiring only a few megabytes with automatic dependency installation.
Rwkv Runner pros
- Only 8MB lightweight executable program
- Full automation with automatic dependency installation
- OpenAI API compatible - works with any ChatGPT client
- One-click model startup and management
- Built-in WebUI for web service sharing
- Built-in model conversion tool
- Built-in download management and remote model inspection
- One-click LoRA finetuning (Windows)
- Pre-set multi-level VRAM configs for all computers
- WebGPU strategy supports AMD and Intel GPUs
- Custom CUDA kernel acceleration for faster inference and less VRAM
- MIDI hardware input with track editing support
- Can be used as OpenAI ChatGPT, GPT-Playground, and Ollama client
- Front-end and back-end separation deployment option
- Multilingual localization with theme switching
- Automatic updates built-in
- Attachment uploads and chat presets supported
- Works as desktop GUI with API support
Rwkv Runner cons
- LoRA finetuning only works on Windows
- Windows Defender may flag as virus requiring exclusion
- Custom CUDA kernel may cause garbled output on some GPUs
- Requires GPU driver upgrade if compatibility issues occur
- Remote service needs request size limits to prevent resource abuse
- Max tokens limited by default (102400) affecting extreme cases
- Backend Python must run separately for server deployment
- Embeddings API results incompatible between v1.4.0 and earlier
- Bluetooth MIDI on Windows requires multiple setup steps
- Sound font files must be manually downloaded and placed
Frequently asked questions about Rwkv Runner
What is RWKV-Runner?
RWKV-Runner is a RWKV management and startup tool that provides full automation with only an 8MB lightweight executable program. It eliminates barriers to using large language models by automating everything and provides an OpenAI API-compatible interface, making every ChatGPT client an RWKV client.
What operating systems does RWKV-Runner support?
RWKV-Runner works on Windows, macOS, and Linux. The RWKV App is available for Android, iOS, PC, Mac, and Linux. One-click LoRA finetuning is Windows Only.
Is RWKV-Runner free?
Yes, RWKV-Runner is completely free and open source under the MIT license. RWKV is fully open source and available for commercial use.
How do I install RWKV-Runner?
Download the executable program (only 8MB) which handles automatic dependency installation. For source deployment: git clone the repository, then run python ./backend-python/main.py for backend service, or python ./backend-python/webui_server.py for frontend, or use --webui flag to start both.
What VRAM configurations are supported?
RWKV-Runner has pre-set multi-level VRAM configs that work well on almost all computers. It supports CUDA acceleration for NVIDIA GPUs and WebGPU strategy for AMD, Intel, and other graphics cards. Custom CUDA kernel acceleration is enabled by default for faster inference and less VRAM consumption.
Can I use RWKV-Runner with ChatGPT clients?
Yes, RWKV-Runner provides an interface compatible with the OpenAI API, which means every ChatGPT client is an RWKV client. After starting the model, open http://127.0.0.1:8000/docs to view API details. You can also use it as an OpenAI ChatGPT, GPT-Playground, and Ollama client by filling in API URL and API Key in Settings.
How do I deploy RWKV-Runner on a server?
You can deploy the backend-python on a server and use the program as a client only by filling in the server address in Settings API URL. For public services, limit request size through API gateway and restrict max_tokens upper limit. See Server Deploy Examples in the repository for detailed instructions.
What is the latest version of RWKV-Runner?
The latest release is v1.9.7 released on August 20, 2025. The project has 98 total releases with automatic updates built-in.
How do I finetune RWKV models with RWKV-Runner?
RWKV-Runner has built-in one-click LoRA Finetuning (Windows Only). Additionally, RWKV-PEFT allows finetuning RWKV with only 9GB VRAM to finetune a 7B model. The tool includes a LoRA Finetune page in the GUI.
What MIDI input options are available?
USB MIDI devices are plug-and-play and selectable in the Composition page. Mac supports Bluetooth MIDI via Bluetooth MIDI Connect app. Windows Bluetooth MIDI requires loopMIDI for virtual device, Bluetooth LE Explorer for pairing, and MIDIberry UWP app to redirect input. Virtual MIDI Controller 3 LE can use regular keyboard as MIDI input.