k/chat
1
Fork 0
Code Issues Pull Requests Wiki Activity

fixes

Browse Source
This commit is contained in:
k
2026-09-21 16:31:55 +02:00
parent b357d89d81
commit 48fb137c13
11 changed files with 78 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/ARCHITECTURE.md
+5
View File
@@ -103,6 +103,11 @@ Models are addressed as `"driver:model"` (split on the first `:`).
- `OllamaDriver`: model discovery via `GET {OLLAMA_HOST}/api/tags`, chat via
`POST /api/chat` with NDJSON streaming.
- `GET /api/models` is served from an in-memory cache in `DriverRegistry`
(`MODEL_CACHE_TTL_MS` = 60 s) so logins don't hit Ollama every time.
Concurrent misses share one in-flight upstream fetch. Admins pass
`?refresh=1` to force a fresh fetch (used by the "Ollama models" view);
the flag is ignored for non-admins. Admin pull/delete invalidate the cache.
- `ApiDriverBase`: shared OpenAI-compatible `/models` + `/chat/completions`
(SSE) implementation; concrete drivers only set base URL, key, and fallback
model list. A driver is enabled only when its `*_ENABLED` flag is set **and**