-
Notifications
You must be signed in to change notification settings - Fork 9
Expand file tree
/
Copy pathmodels.example.json
More file actions
32 lines (32 loc) · 1.48 KB
/
Copy pathmodels.example.json
File metadata and controls
32 lines (32 loc) · 1.48 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
{
"_readme": [
"Optional per-instance provider config. Crow does NOT ship a models.json:",
"on a fresh install the providers table starts empty and you add providers",
"in the dashboard (Settings -> AI Models -> Providers) or by installing a",
"model bundle from Extensions, which registers its provider automatically.",
"",
"Advanced/recovery use only: copy this file to ./models.json or",
"./config/models.json (both are gitignored) and edit it. On first boot with",
"an EMPTY providers table its entries seed the table; afterwards the",
"Settings -> AI Models -> Providers -> 'Sync bundle providers' button",
"re-upserts entries you own. It is also the disaster-recovery fallback the",
"gateway reads if the database is unreachable.",
"",
"Each key under 'providers' is a provider id. baseUrl points at an",
"OpenAI-compatible endpoint. 'models' lists the model ids that endpoint",
"serves. Optional fields: apiKey, host ('local'|'cloud'), bundleId,",
"description, mutexGroup, alwaysResident, defaultMember.",
"",
"The CROW_MODELS_JSON env var (colon-separated paths) overrides where",
"these files are looked for; set it to an empty string to ignore them."
],
"providers": {
"example-local": {
"baseUrl": "http://localhost:8080/v1",
"description": "Example: a llama.cpp/vLLM/ollama-style server on this machine",
"models": [
{ "id": "example-model-id", "name": "Example Model" }
]
}
}
}