AI: OpenAI-compatible provider now uses /v1/chat/completions (not the legacy /v1/completions) so Nvidia NIM and similar work; tolerate a trailing /v1 in the base URL (#96)

This commit is contained in:
Yuri Karamian committed 2026-06-11 22:36:03 +02:00
1 parent d9cb615f1f
commit fb4c5748a8
2 files changed
+12 -166

No files matched your search

+1 -1
View File
@@ -1313,7 +1313,7 @@
oninput={(e) => { _openaiCompatibleBaseUrl = (e.target as HTMLInputElement).value; }}
onblur={saveAiSettings}
/>
<p class="setting-hint">Base URL for any OpenAI-compatible server (LM Studio, LocalAI, vLLM, etc.). Uses the <code>v1/completions</code> endpoint.</p>
<p class="setting-hint">Base URL for any OpenAI-compatible server (LM Studio, LocalAI, vLLM, Nvidia NIM, etc.). Enter the server root; the app appends <code>/v1/chat/completions</code> (a trailing <code>/v1</code> is fine too).</p>
</div>
<div class="settings-section">
<h3>API Key <span style="font-weight: 400; text-transform: none; font-size: 11px;">(optional)</span></h3>