pin model permanently warm (keep_alive -1) — no cold loads

64585736a20c9168ce662fef679db56f20f299aa · 2026-08-19 11:02:48 -0700 · steve

Files touched

M server.js

Diff

commit 64585736a20c9168ce662fef679db56f20f299aa
Author: steve <steve@designerwallcoverings.com>
Date:   Wed Aug 19 11:02:48 2026 -0700

    pin model permanently warm (keep_alive -1) — no cold loads
---
 server.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/server.js b/server.js
index ce00b01..4c74838 100644
--- a/server.js
+++ b/server.js
@@ -14,7 +14,7 @@ const app = express();
 const PORT = process.env.PORT || 9821;
 const OLLAMA = process.env.OLLAMA_URL || 'http://127.0.0.1:11434';
 const MODEL = process.env.MODEL || 'qwen3.8-27b-heretic';
-const KEEP_ALIVE = process.env.KEEP_ALIVE || '30m'; // keep model warm to avoid ~2min cold loads
+const KEEP_ALIVE = process.env.KEEP_ALIVE || '-1'; // -1 = never unload (pinned warm, instant responses)
 
 // ---- Basic auth gate — multi-user ------------------------------------------
 // USERS env = "user1:pass1,user2:pass2". Falls back to admin/DW2024!.

← 78d2199 viewer: motion-graphics thinking feedback (pulsing orb, elap · back to Qwen38 Viewer · send keep_alive as integer -1 (valid 'keep forever') 411e696 →