FOR CALLERS
Send a prompt. Pay per token.
Open-weights models, on tap. One funded key; the cheapest awake machine answers; the prompt stays on that machine.
0 computers on · 0 models · 0 tok/s
Ling 3.0 Flash
$0.063 / 1M listed
FOR HOSTS
Run a model. Keep 80%.
This tab claims jobs for your machine. Payouts settle in shielded Zcash.
Hosts keep 80% of every job.
Nothing awake yet. The listed floor is $1.00 / 1M.
At $1.00 / 1M and 30 tok/s, one busy hour pays about $0.09. Estimate, not a promise.
Plugs into
1/3
1Check your engine
Pick what is already running. Then check that this page can reach it. A failed check means callers would fail too.
Choose a network model
Three pinned catalog entries to start. A model name is a declaration, not proof of weights. Your current model can still serve if it is not listed.
vLLM or another compatible OpenAI endpoint. Llama 3.1 Community License; review terms before hosting.
Source revision f0656779 ↗Use the alias
Dolphin3.0-Llama3.1-8Bonly for this build. Other quantizations need review; memory figures exclude runtime overhead.Restart the server so this page may call it. Do not put a key on it.
Copy the name from ollama list, tag included. qwen2.5:7b is not qwen2.5.
OLLAMA_ORIGINS=https://owterminal.com ollama serve
2Join the pool
3You're live