diff --git a/.github/workflows/interactive.yml b/.github/workflows/interactive.yml
index 279603cbca35..716c79ac2627 100644
--- a/.github/workflows/interactive.yml
+++ b/.github/workflows/interactive.yml
@@ -605,3 +605,15 @@ jobs:
         SCHEMA_FILE=${GITHUB_WORKSPACE}/flex/tests/rt_mutable_graph/movie_schema_test.yaml
         BULK_LOAD_FILE=${GITHUB_WORKSPACE}/flex/tests/rt_mutable_graph/movie_import_test.yaml
         GLOG_v=10 ./bin/bulk_loader -g ${SCHEMA_FILE} -l ${BULK_LOAD_FILE} -d /tmp/csr-data-dir/
+    
+    - name: Test graph loading with different delimiter
+      env:
+        GS_TEST_DIR: ${{ github.workspace }}/gstest/
+        FLEX_DATA_DIR: ${{ github.workspace }}/gstest/flex/modern_graph_tab_delimiter/
+      run: |
+        rm -rf /tmp/csr-data-dir/
+        cd ${GITHUB_WORKSPACE}/flex/build/
+        SCHEMA_FILE=${GITHUB_WORKSPACE}/flex/interactive/examples/modern_graph/graph.yaml
+        BULK_LOAD_FILE=${GITHUB_WORKSPACE}/flex/interactive/examples/modern_graph/bulk_load.yaml
+        sed -i 's/|/\\t/g' ${BULK_LOAD_FILE}
+        GLOG_v=10 ./bin/bulk_loader -g ${SCHEMA_FILE} -l ${BULK_LOAD_FILE} -d /tmp/csr-data-dir/
diff --git a/docs/flex/interactive/data_import.md b/docs/flex/interactive/data_import.md
index 7b7b454b5aa6..90d71eb350bc 100644
--- a/docs/flex/interactive/data_import.md
+++ b/docs/flex/interactive/data_import.md
@@ -227,7 +227,7 @@ The table below offers a detailed breakdown of each configuration item. In this
 | loading_config.scheme | file | The source of input data. Currently only `file` and `odps` are supported | No |
 | loading_config.format    | N/A     | The format of the raw data in CSV    |  Yes   |
 | loading_config.format.metadata    | N/A    | Mainly for configuring the options for reading CSV   |  Yes   |
-| loading_config.format.metadata.delimiter | '\|' | Delimiter used to split a row of data | Yes | 
+| loading_config.format.metadata.delimiter | '|' | Delimiter used to split a row of data, escaped char are also supported, i.e. '\t' | Yes | 
 | loading_config.format.metadata.header_row | true | Indicate if the first row should be used as the header | No | 
 | loading_config.format.metadata.quoting | false | Whether quoting is used | No |
 | loading_config.format.metadata.quote_char | '\"' | Quoting character (if `quoting` is true) | No |
diff --git a/flex/storages/rt_mutable_graph/loader/csv_fragment_loader.cc b/flex/storages/rt_mutable_graph/loader/csv_fragment_loader.cc
index ef2f9412b0dc..682a6a460e3b 100644
--- a/flex/storages/rt_mutable_graph/loader/csv_fragment_loader.cc
+++ b/flex/storages/rt_mutable_graph/loader/csv_fragment_loader.cc
@@ -158,10 +158,25 @@ static std::vector<std::string> read_header(
 static void put_delimiter_option(const LoadingConfig& loading_config,
                                  arrow::csv::ParseOptions& parse_options) {
   auto delimiter_str = loading_config.GetDelimiter();
-  if (delimiter_str.size() != 1) {
-    LOG(FATAL) << "Delimiter should be a single character";
+  if (delimiter_str.size() != 1 && delimiter_str[0] != '\\') {
+    LOG(FATAL) << "Delimiter should be a single character, or a escape "
+                  "character, like '\\t'";
+  }
+  if (delimiter_str[0] == '\\') {
+    if (delimiter_str.size() != 2) {
+      LOG(FATAL) << "Delimiter should be a single character";
+    }
+    // escape the special character
+    switch (delimiter_str[1]) {
+    case 't':
+      parse_options.delimiter = '\t';
+      break;
+    default:
+      LOG(FATAL) << "Unsupported escape character: " << delimiter_str[1];
+    }
+  } else {
+    parse_options.delimiter = delimiter_str[0];
   }
-  parse_options.delimiter = delimiter_str[0];
 }
 
 static bool put_skip_rows_option(const LoadingConfig& loading_config,