diff --git a/src/services/code-index/constants/index.ts b/src/services/code-index/constants/index.ts
index 6f0e0fe7e6..1beb140992 100644
--- a/src/services/code-index/constants/index.ts
+++ b/src/services/code-index/constants/index.ts
@@ -24,6 +24,7 @@ export const MAX_PENDING_BATCHES = 20 // Maximum number of batches to accumulate
 
 /**OpenAI Embedder */
 export const MAX_BATCH_TOKENS = 100000
+export const MAX_BATCH_ITEMS = 32 // Maximum number of items per embedding API call (provider limit, fixes #335)
 export const MAX_ITEM_TOKENS = 8191
 export const BATCH_PROCESSING_CONCURRENCY = 10
 
diff --git a/src/services/code-index/embedders/bedrock.ts b/src/services/code-index/embedders/bedrock.ts
index b03d062cb7..a229823962 100644
--- a/src/services/code-index/embedders/bedrock.ts
+++ b/src/services/code-index/embedders/bedrock.ts
@@ -3,6 +3,7 @@ import { fromIni, fromNodeProviderChain } from "@aws-sdk/credential-providers"
 import { IEmbedder, EmbeddingResponse, EmbedderInfo } from "../interfaces"
 import {
 	MAX_BATCH_TOKENS,
+	MAX_BATCH_ITEMS,
 	MAX_ITEM_TOKENS,
 	MAX_BATCH_RETRIES as MAX_RETRIES,
 	INITIAL_RETRY_DELAY_MS as INITIAL_DELAY_MS,
@@ -83,7 +84,7 @@ export class BedrockEmbedder implements IEmbedder {
 					continue
 				}
 
-				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS) {
+				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS && currentBatch.length < MAX_BATCH_ITEMS) {
 					currentBatch.push(text)
 					currentBatchTokens += itemTokens
 					processedIndices.push(i)
diff --git a/src/services/code-index/embedders/openai-compatible.ts b/src/services/code-index/embedders/openai-compatible.ts
index 6eaf2b6c2c..6ae82ebf48 100644
--- a/src/services/code-index/embedders/openai-compatible.ts
+++ b/src/services/code-index/embedders/openai-compatible.ts
@@ -2,6 +2,7 @@ import { OpenAI } from "openai"
 import { IEmbedder, EmbeddingResponse, EmbedderInfo } from "../interfaces/embedder"
 import {
 	MAX_BATCH_TOKENS,
+	MAX_BATCH_ITEMS,
 	MAX_ITEM_TOKENS,
 	MAX_BATCH_RETRIES as MAX_RETRIES,
 	INITIAL_RETRY_DELAY_MS as INITIAL_DELAY_MS,
@@ -144,7 +145,7 @@ export class OpenAICompatibleEmbedder implements IEmbedder {
 					continue
 				}
 
-				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS) {
+				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS && currentBatch.length < MAX_BATCH_ITEMS) {
 					currentBatch.push(text)
 					currentBatchTokens += itemTokens
 					processedIndices.push(i)
diff --git a/src/services/code-index/embedders/openai.ts b/src/services/code-index/embedders/openai.ts
index b993e280d9..ed8c52943a 100644
--- a/src/services/code-index/embedders/openai.ts
+++ b/src/services/code-index/embedders/openai.ts
@@ -4,6 +4,7 @@ import { ApiHandlerOptions } from "../../../shared/api"
 import { IEmbedder, EmbeddingResponse, EmbedderInfo } from "../interfaces"
 import {
 	MAX_BATCH_TOKENS,
+	MAX_BATCH_ITEMS,
 	MAX_ITEM_TOKENS,
 	MAX_BATCH_RETRIES as MAX_RETRIES,
 	INITIAL_RETRY_DELAY_MS as INITIAL_DELAY_MS,
@@ -100,7 +101,7 @@ export class OpenAiEmbedder extends OpenAiNativeHandler implements IEmbedder {
 					continue
 				}
 
-				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS) {
+				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS && currentBatch.length < MAX_BATCH_ITEMS) {
 					currentBatch.push(text)
 					currentBatchTokens += itemTokens
 					processedIndices.push(i)
diff --git a/src/services/code-index/embedders/openrouter.ts b/src/services/code-index/embedders/openrouter.ts
index d98aaeeeb5..b2851d2be7 100644
--- a/src/services/code-index/embedders/openrouter.ts
+++ b/src/services/code-index/embedders/openrouter.ts
@@ -2,6 +2,7 @@ import { OpenAI } from "openai"
 import { IEmbedder, EmbeddingResponse, EmbedderInfo } from "../interfaces/embedder"
 import {
 	MAX_BATCH_TOKENS,
+	MAX_BATCH_ITEMS,
 	MAX_ITEM_TOKENS,
 	MAX_BATCH_RETRIES as MAX_RETRIES,
 	INITIAL_RETRY_DELAY_MS as INITIAL_DELAY_MS,
@@ -148,7 +149,7 @@ export class OpenRouterEmbedder implements IEmbedder {
 					continue
 				}
 
-				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS) {
+				if (currentBatchTokens + itemTokens <= MAX_BATCH_TOKENS && currentBatch.length < MAX_BATCH_ITEMS) {
 					currentBatch.push(text)
 					currentBatchTokens += itemTokens
 					processedIndices.push(i)