Your private mesh
Mesh LLM lets you build private personal AI, using open models. Pool machines across your home, office, or friends, no cloud needed. Learn more → · GitHub
Mesh has mixed versions
Detected 7 distinct mesh-llm versions: 0.65.0-rc2, 0.60.2, 0.58.0, 0.64.0, 0.62.1, 0.55.1, 0.56.0. Functionality may vary between nodes.
Node ID
a3cd1b06a7Client
Owner
UnsignedUnsigned
Active Models
6
Mesh VRAM
381.5 GB
Nodes
7+6 clients
Inflight
0
Mesh Overview
Fullscreen
6834941b7eede8
Client
Model Catalog
FilterAll
GLM-4.7-Flash-Q4_K_M
🧩
GLM-4.7-Flash-Q4_K_M
Warm
1 node18.0 GB
Hermes-2-Pro-Mistral-7B-Q4_K_M
Hermes-2-Pro-Mistral-7B-Q4_K_M
Warm
1 node4.4 GB
MiniMax-M2.5-Q4_K_M
MiniMax-M2.5-Q4_K_M
Warm
1 node138.0 GB
Qwen3-4B-Q4_K_M
Qwen3-4B-Q4_K_M
Warm
1 node2.5 GB
Qwen3-8B-Q4_K_M
Qwen3-8B-Q4_K_M
Warm
1 node5.0 GB
Qwen3.5-9B-Vision-Q4_K_M
Qwen3.5-9B-Q4_K_M
Warm
1 node5.8 GB
Qwen3.5-27B-Q4_K_M
Qwen3.5-27B-Q4_K_M
Cold
0 nodes17.0 GB
Connected Peers
| ID | Role | Version | Status | Model | Latency | VRAM | Share |
|---|---|---|---|---|---|---|---|
| 1cb2046c44 | Host | 0.65.0-rc2 | Serving | MiniMax-M2.5 | <1 ms | 256.0 GB | 67% |
| 79af94d8be | Host | 0.62.1 | Serving | GLM-4.7-Flash | <1 ms | 38.7 GB | 10% |
| a778f419f1 | Worker | 0.64.0 | Serving | Qwen3.5-27B | <1 ms | 38.7 GB | 10% |
| 3bb8566e6d | Host | 0.65.0-rc2 | Serving | Qwen3.5-9B | <1 ms | 16.0 GB | 4% |
| 5714ff65c4 | Host | 0.64.0 | Serving | Qwen3-4B | <1 ms | 12.9 GB | 3% |
| 5fb0a7fa5a | Host | 0.56.0 | Serving | Hermes-2-Pro-Mistral-7B | <1 ms | 12.9 GB | 3% |
| 1a3457494a | Host | 0.60.2 | Serving | Qwen3-8B | <1 ms | 6.4 GB | 2% |
| 03a5aac048 | Client | 0.64.0 | Client | idle | <1 ms | n/a | n/a |
| 1230138ce8 | Client | 0.58.0 | Client | idle | — | n/a | n/a |
| 7d6e5ac6ea | Client | 0.64.0 | Client | idle | <1 ms | n/a | n/a |
| bfc5ab60e4 | Client | 0.62.1 | Client | idle | <1 ms | n/a | n/a |
| f5779680e6 | Client | 0.55.1 | Client | idle | <1 ms | n/a | n/a |
Connect
Run mesh-llm on your machine to get a local OpenAI-compatible API and contribute compute to the mesh.
- Install
- Run
mesh-llm --auto --join eyJpZCI6ImEzY2QxYjA2YTc3MjA1OWVlYTNkMTY5OTYxYzQyNjc0MzIwZjAxNjdlYjdkMjU0OGE2ZDlmNzQzNDJiMjhjODEiLCJhZGRycyI6W3siUmVsYXkiOiJodHRwczovL2FwczEtMS5yZWxheS5taWNoYWVsbmVhbGUubWVzaC1sbG0uaXJvaC5saW5rLi8ifSx7IklwIjoiMTcyLjE5LjEyLjU4OjQ1NzA4In0seyJJcCI6IjE3Mi4xOS4xMi41OTo0NTcwOCJ9LHsiSXAiOiJbMjYwNTo0YzQwOjMzOjE0NTg6MDo5NTMxOmMzZGE6MV06MzUzODQifV19
This auto-selects a model for your hardware, joins the mesh, and serves a local API at http://127.0.0.1:9337/v1.