Llama-Server Web Client (PHP)

Connection

PHP ?

Admin Off

DB ?

Models ?

CSRF: 89db42856bf9db6d12cbdb0f418b316c

Available models (router mode)

Unknown

Model list comes from GET /models (router mode) or GET /v1/models (single-model).

Upload files to embed into MariaDB Vector

Embeddings are generated via POST /v1/embeddings and stored using MariaDB VECTOR + VECTOR INDEX.

If model switching doesn't work, start llama-server in router mode (no -m argument).
Streaming uses SSE. If you see buffering, disable proxy buffering (e.g., Nginx X-Accel-Buffering).
For RAG, run DB Setup once; then embed files; then enable RAG toggle in chat options.

Message

Streaming RAG

Temperature

Max tokens

RAG top-k

System prompt (optional)