OpenLLM runs open-source language models as cloud API endpoints compatible with OpenAI clients.
Open source
21 alternativesNobodyWho
NobodyWho runs language models locally through an inference engine designed for efficient operation across devices.
About NobodyWho
NobodyWho is an open-source inference engine for running large language models (LLMs) locally on a device. It is built on llama.cpp and runs text, vision and embedding models, along with speech-to-text and text-to-speech. The homepage states that it runs offline and needs no API keys. It is available through bindings for Kotlin, Swift, React Native/Expo, Flutter, Python and Godot.
Memory is the main practical limit. The homepage says models need roughly 1.5 times the model file size in free RAM, or about 2 times on a busy machine. Desktop support covers 64-bit Linux, macOS and Windows (x86_64 only). Godot has no iOS export, Windows ARM64 is not supported, and there is no web export.
Key features
- Runs LLMs locally with GPU acceleration via Vulkan or Metal, falling back to CPU
- Loads GGUF models from Hugging Face or any URL
- Tool calling with grammars generated automatically from function signatures
- Multimodal input with image and audio information for LLMs
- Speech-to-text with Whisper and voice activity detection with Silero
- Text-to-speech that synthesizes local WAV audio with Kokoro, Pocket TTS and Supertonic
Best for: Developers who want to run LLMs locally inside desktop or mobile apps, or in Python projects, without relying on an external API.
Good for
- Runs locally and offline, and the homepage states that no API keys are needed
- Uses GPU acceleration through Vulkan or Metal when available, with CPU fallback
- Open source, with a public repository on GitHub
Watch out for
- Needs roughly 1.5 times the model file size in free RAM, or about 2 times on a busy machine
- Godot has no iOS export, Windows ARM64 is not supported, and there is no web export (tracked in issue #111)
- On discrete GPUs, models larger than VRAM still run but more slowly, with remaining layers on CPU
NobodyWho at a glance
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Open source
- Yes · Source code
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
NobodyWho alternatives
Tools and sites you could use instead, ranked by how closely they match.
GPUStack manages GPU clusters for serving AI models and provides on demand GPU instances accessible through SSH.
Open source
21 alternatives
Rapid-MLX is an open-source LLM inference server for Apple Silicon, built on MLX, compatible with OpenAI and Anthropic APIs, focused on reliable tool calling for coding agents.
Open source
14 alternatives
LLMKube runs self hosted LLM inference across heterogeneous GPU fleets using Kubernetes and several model runtimes.
Open source
22 alternatives
PowerInfer supports serving large language models for local deployment and inference.
Open source
23 alternatives
Sonar provides large-scale inference for language models through the project hosted in this repository.
Open source
23 alternativesPopular in Model Serving & Inference
Well-known listings from the same topic.
MAI Playgroundplayground.microsoft.ai/chatBigModelbigmodel.cnkimi-k2.6platform.kimi.ai/docs/pricing/chat-k26Jinagithub.com/jina-ai/jinaFluxgithub.com/black-forest-labs/fluxprivateGPTgithub.com/imartinez/privateGPT
NobodyWho FAQ
Which platforms are supported?
Desktop means 64-bit Linux, macOS and Windows (x86_64 only). Bindings are listed for Kotlin (desktop, Android), Swift (macOS, iOS, visionOS, watchOS), React Native/Expo (desktop, Android, iOS), Flutter (desktop, Android, iOS), Python (desktop) and Godot (desktop, Android).
How much memory does a model need?
The homepage says roughly 1.5 times the model file in free RAM, or 2 times on a machine that is already busy. An 8 GB machine is described as a comfortable floor for models up to about 2 GB, and 16 GB or more for larger models.
What happens if a model is larger than the GPU's memory?
NobodyWho offloads as many layers as fit in free VRAM and runs the rest on the CPU. The model still works but runs more slowly. If very little fits, it skips the GPU and runs on CPU.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.