Ollama

Ollama Chat

Ollama base URL is https://api.siliconnetwork.io/ollama. Discovery needs no key; chat does, so use a client that can set an Authorization header.

curl -X POST https://api.siliconnetwork.io/ollama/api/chat \
  -H "Authorization: Bearer sn_live_..." \
  -H "Content-Type: application/json" \
  -d '{
  "model": "llama-3.2-3b",
  "messages": [
    {
      "role": "user",
      "content": "Say hello in one line."
    }
  ],
  "stream": false
}'

Responses: 200 Successful Response, 422 Validation Error

Ollama Generate

curl -X POST https://api.siliconnetwork.io/ollama/api/generate \
  -H "Authorization: Bearer sn_live_..." \
  -H "Content-Type: application/json" \
  -d '{
  "model": "llama-3.2-3b",
  "prompt": "Say hello in one line.",
  "stream": false
}'

Responses: 200 Successful Response, 422 Validation Error

Ollama Root

Responses: 200 Successful Response

Ollama Root

Responses: 200 Successful Response

Ollama Root

Responses: 200 Successful Response

Ollama Root

Responses: 200 Successful Response

Ollama Version

Clients only check that this answers and parses; the value mirrors the engine's.

Responses: 200 Successful Response

Ollama Tags

The models an Ollama client may actually call, synthesized from what is LOADED. Deliberately NOT a proxy of the engine's own /ollama/api/tags: that lists every model whose weights have been downloaded, and almost none of those have a running instance. Proxying it would reproduce the 404 inside every client's model picker; a menu of names, one of which works.

Lists the models loaded right now, so an Ollama client's picker only shows ids that serve.

Responses: 200 Successful Response

Ollama Ps

Running models. The same list as /api/tags, because at the edge only a loaded model is callable, so 'installed' and 'running' cannot differ here.

Responses: 200 Successful Response