From 7ab9a7fd2d308ceb653d3b91033e20abb4215c3d Mon Sep 17 00:00:00 2001
From: Developer <developer@discord-automod.local>
Date: Fri, 31 Jul 2026 20:14:18 +0700
Subject: [PATCH] =?UTF-8?q?fix(automod):=20force=20float=20encoding=20for?=
 =?UTF-8?q?=20embeddings=20=E2=80=94=20Nvidia=20models=20reject=20base64?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

OpenAI SDK v6 defaults to encoding_format=base64; llama-nemotron-embed
(Nvidia-backed) returns 400 'do not support base64'. Semantic cache was
silently disabled in prod. encoding_format: 'float' fixes it.
---
 .../src/modules/ai-moderation/embeddingClient.ts               | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/services/discord-gateway/src/modules/ai-moderation/embeddingClient.ts b/services/discord-gateway/src/modules/ai-moderation/embeddingClient.ts
index 9ebb436..29373df 100644
--- a/services/discord-gateway/src/modules/ai-moderation/embeddingClient.ts
+++ b/services/discord-gateway/src/modules/ai-moderation/embeddingClient.ts
@@ -61,6 +61,9 @@ export async function embedTexts(texts: string[]): Promise<number[][] | null> {
     const response = await client.embeddings.create({
       model: config.AI_LLM_EMBEDDING_MODEL as string,
       input: texts,
+      // OpenAI SDK v6 defaults to base64; Nvidia-backed embedding models
+      // (e.g. llama-nemotron-embed) reject it with 400. Always float.
+      encoding_format: "float",
     });
     return response.data.map((item) => item.embedding);
   } catch (error) {