Skip to content

RAGFlow is an agent-native database: a Retrieval-Augmented Generation (RAG) engine that grew into context infrastructure.

feat: add batch_size to all embedding model configs (#17877)

jay777212e2d12d26225
diff --git a/conf/models/302ai.json b/conf/models/302ai.json
index 1ac370c6a2..d1e11c465f 100644
--- a/conf/models/302ai.json
+++ b/conf/models/302ai.json
@@ -293,7 +293,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-reranker-v2-base-multilingual",
diff --git a/conf/models/aliyun.json b/conf/models/aliyun.json
index 0f4c845149..01c2e92349 100644
--- a/conf/models/aliyun.json
+++ b/conf/models/aliyun.json
@@ -30,7 +30,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 10
},
{
"name": "qwen3-vl-plus",
@@ -67,7 +68,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 10
},
{
"name": "qwen3-rerank",
diff --git a/conf/models/astraflow.json b/conf/models/astraflow.json
index 6b33a8fd0e..97f246edfb 100644
--- a/conf/models/astraflow.json
+++ b/conf/models/astraflow.json
@@ -18,7 +18,8 @@
"max_tokens": 16384,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 512
},
{
"name": "bge-reranker-v2-m3",
diff --git a/conf/models/baichuan.json b/conf/models/baichuan.json
index cfe927d32f..c69be2ba1b 100644
--- a/conf/models/baichuan.json
+++ b/conf/models/baichuan.json
@@ -124,7 +124,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16
}
]
}
diff --git a/conf/models/baidu.json b/conf/models/baidu.json
index 726ffddaa1..e86119b9b1 100644
--- a/conf/models/baidu.json
+++ b/conf/models/baidu.json
@@ -88,7 +88,8 @@
"max_tokens": 384,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16
},
{
"name": "qwen3-reranker-4b",
diff --git a/conf/models/bedrock.json b/conf/models/bedrock.json
index 69a94e7eac..954e92f258 100644
--- a/conf/models/bedrock.json
+++ b/conf/models/bedrock.json
@@ -132,35 +132,40 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1
},
{
"name": "amazon.titan-embed-text-v1",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1
},
{
"name": "cohere.embed-english-v3",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "cohere.embed-multilingual-v3",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "cohere.embed-v4:0",
"max_tokens": 128000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
}
]
}
diff --git a/conf/models/cohere.json b/conf/models/cohere.json
index 244c9a6b2b..cd943f3cb6 100644
--- a/conf/models/cohere.json
+++ b/conf/models/cohere.json
@@ -117,35 +117,40 @@
"max_tokens": 131072,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "embed-english-v3.0",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "embed-english-light-v3.0",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "embed-multilingual-v3.0",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "embed-multilingual-light-v3.0",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 96
},
{
"name": "cohere-transcribe-03-2026",
diff --git a/conf/models/cometapi.json b/conf/models/cometapi.json
index 62957a4d71..e316de475f 100644
--- a/conf/models/cometapi.json
+++ b/conf/models/cometapi.json
@@ -124,21 +124,24 @@
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "text-embedding-3-large",
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "text-embedding-ada-002",
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "whisper-1",
diff --git a/conf/models/deepinfra.json b/conf/models/deepinfra.json
index 4269ba3e0a..aa4387a0d7 100644
--- a/conf/models/deepinfra.json
+++ b/conf/models/deepinfra.json
@@ -34,7 +34,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1024
},
{
"name": "hexgrad/Kokoro-82M",
diff --git a/conf/models/gitee.json b/conf/models/gitee.json
index 7ae3ea2dda..035be6ce1d 100644
--- a/conf/models/gitee.json
+++ b/conf/models/gitee.json
@@ -63,7 +63,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "GOT-OCR2_0",
@@ -87,7 +88,8 @@
"name": "jina-clip-v2",
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "HunyuanOCR",
diff --git a/conf/models/google.json b/conf/models/google.json
index 86ab91d47d..f3d14de163 100644
--- a/conf/models/google.json
+++ b/conf/models/google.json
@@ -29,7 +29,8 @@
"max_tokens": 2048,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 100
}
],
"features": {
diff --git a/conf/models/greenpt.json b/conf/models/greenpt.json
index c9b1a48da9..9808b4bc57 100644
--- a/conf/models/greenpt.json
+++ b/conf/models/greenpt.json
@@ -43,7 +43,8 @@
],
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "green-rerank",
diff --git a/conf/models/huaweicloud.json b/conf/models/huaweicloud.json
index 2a5f5d802a..310a85f6f2 100755
--- a/conf/models/huaweicloud.json
+++ b/conf/models/huaweicloud.json
@@ -218,7 +218,8 @@
"max_tokens": 8000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "bge-reranker-v2-m3",
diff --git a/conf/models/hunyuan.json b/conf/models/hunyuan.json
index a1ea4bcef1..2a2bcfe186 100644
--- a/conf/models/hunyuan.json
+++ b/conf/models/hunyuan.json
@@ -364,28 +364,32 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 50
},
{
"name": "kinfra-text-embedding-4b",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 50
},
{
"name": "kinfra-vl-embedding-2b",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 50
},
{
"name": "kinfra-vl-embedding-8b",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 50
}
]
}
diff --git a/conf/models/jiekouai.json b/conf/models/jiekouai.json
index 9deac93df9..a4c2320eb4 100644
--- a/conf/models/jiekouai.json
+++ b/conf/models/jiekouai.json
@@ -128,7 +128,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
}
]
}
diff --git a/conf/models/jina.json b/conf/models/jina.json
index ca934930d7..28f2991ebb 100644
--- a/conf/models/jina.json
+++ b/conf/models/jina.json
@@ -56,56 +56,64 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-embeddings-v4",
"max_tokens": 32768,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-embeddings-v5-text-small",
"max_tokens": 32768,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-embeddings-v5-text-nano",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-embeddings-v5-omni-small",
"max_tokens": 32768,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-embeddings-v5-omni-nano",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-clip-v2",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
},
{
"name": "jina-embeddings-v2-base-en",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16384
}
]
}
diff --git a/conf/models/mistral.json b/conf/models/mistral.json
index f140e337e7..616b62db54 100644
--- a/conf/models/mistral.json
+++ b/conf/models/mistral.json
@@ -174,7 +174,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16
},
{
"name": "mistral-ocr-2512",
diff --git a/conf/models/n1n.json b/conf/models/n1n.json
index ce5f011e7d..8a34da2a28 100644
--- a/conf/models/n1n.json
+++ b/conf/models/n1n.json
@@ -119,21 +119,24 @@
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "text-embedding-3-large",
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "text-embedding-ada-002",
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "BAAI/bge-reranker-v2-m3",
diff --git a/conf/models/novita.json b/conf/models/novita.json
index bddaaae858..d2a5fa7e18 100644
--- a/conf/models/novita.json
+++ b/conf/models/novita.json
@@ -94,7 +94,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "baai/bge-reranker-v2-m3",
diff --git a/conf/models/nvidia.json b/conf/models/nvidia.json
index 9d4ee14d64..05c7fb39b4 100644
--- a/conf/models/nvidia.json
+++ b/conf/models/nvidia.json
@@ -234,7 +234,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "nvidia/nemotron-3-nano-30b-a3b",
@@ -315,14 +316,16 @@
"max_tokens": 32768,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "nvidia/nv-embedcode-7b-v1",
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "nvidia/nvidia-nemotron-nano-9b-v2",
diff --git a/conf/models/openai.json b/conf/models/openai.json
index 1f3e2ffd73..0c15f86ab8 100644
--- a/conf/models/openai.json
+++ b/conf/models/openai.json
@@ -255,21 +255,24 @@
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "text-embedding-3-small",
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "text-embedding-3-large",
"max_tokens": 8191,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "whisper-1",
diff --git a/conf/models/perplexity.json b/conf/models/perplexity.json
index 793d3e2ebc..961976ec19 100644
--- a/conf/models/perplexity.json
+++ b/conf/models/perplexity.json
@@ -51,14 +51,16 @@
"max_tokens": 32000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 512
},
{
"name": "pplx-embed-v1-4b",
"max_tokens": 32000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 512
}
]
}
diff --git a/conf/models/ppio.json b/conf/models/ppio.json
index 5886b2acc7..6ce4441328 100644
--- a/conf/models/ppio.json
+++ b/conf/models/ppio.json
@@ -247,19 +247,22 @@
"name": "qwen/qwen3-embedding-8b",
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "baai/bge-m3",
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "qwen/qwen3-embedding-0.6b",
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "baai/bge-reranker-v2-m3",
diff --git a/conf/models/replicate.json b/conf/models/replicate.json
index ca60f57537..103eb9b2e6 100644
--- a/conf/models/replicate.json
+++ b/conf/models/replicate.json
@@ -47,14 +47,16 @@
"max_tokens": 384,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16
},
{
"name": "ibm-granite/granite-embedding-278m-multilingual:1f76d42a05f120e12272746d5a2d86b525c13420773f795a4cbef9117d8685f1",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 16
}
]
}
diff --git a/conf/models/siliconflow.json b/conf/models/siliconflow.json
index fb1c07d1c5..1af3894b72 100644
--- a/conf/models/siliconflow.json
+++ b/conf/models/siliconflow.json
@@ -120,7 +120,8 @@
"max_tokens": 8192,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 32
},
{
"name": "BAAI/bge-m3",
@@ -129,7 +130,8 @@
"embedding"
],
"max_dimension": 1024,
- "dimensions": []
+ "dimensions": [],
+ "batch_size": 32
},
{
"name": "fnlp/MOSS-TTSD-v0.5",
diff --git a/conf/models/togetherai.json b/conf/models/togetherai.json
index 690ec34c1d..e42c4945e5 100644
--- a/conf/models/togetherai.json
+++ b/conf/models/togetherai.json
@@ -51,21 +51,24 @@
"max_tokens": 514,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "BAAI/bge-large-en-v1.5",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "BAAI/bge-base-en-v1.5",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 2048
},
{
"name": "mixedbread-ai/mxbai-rerank-large-v2",
diff --git a/conf/models/upstage.json b/conf/models/upstage.json
index ac2a5d223f..2162499780 100644
--- a/conf/models/upstage.json
+++ b/conf/models/upstage.json
@@ -59,14 +59,16 @@
"max_tokens": 2000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 100
},
{
"name": "solar-embedding-1-large-passage",
"max_tokens": 2000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 100
}
]
}
diff --git a/conf/models/volcengine.json b/conf/models/volcengine.json
index 6da91e021b..49743b464c 100644
--- a/conf/models/volcengine.json
+++ b/conf/models/volcengine.json
@@ -32,7 +32,8 @@
"max_tokens": 131072,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 10
}
]
}
diff --git a/conf/models/voyage.json b/conf/models/voyage.json
index 2873ff4539..fd4b29b1a3 100644
--- a/conf/models/voyage.json
+++ b/conf/models/voyage.json
@@ -14,63 +14,72 @@
"max_tokens": 32000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-4",
"max_tokens": 32000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-4-lite",
"max_tokens": 32000,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-3.5",
"max_tokens": 327680,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-3.5-lite",
"max_tokens": 1048576,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-3-large",
"max_tokens": 122880,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-code-3",
"max_tokens": 122880,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-law-2",
"max_tokens": 122880,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "voyage-finance-2",
"max_tokens": 122880,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 1000
},
{
"name": "rerank-2.5",
diff --git a/conf/models/zhipu-ai.json b/conf/models/zhipu-ai.json
index c55522420e..4f92c6ec4e 100644
--- a/conf/models/zhipu-ai.json
+++ b/conf/models/zhipu-ai.json
@@ -356,14 +356,16 @@
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 512
},
{
"name": "embedding-3",
"max_tokens": 512,
"model_types": [
"embedding"
- ]
+ ],
+ "batch_size": 512
},
{
"name": "glm-asr-2512",