diff --git a/pipeline/clean/opuscleaner/configs/en-ja/default.filters.json b/pipeline/clean/opuscleaner/configs/en-ja/default.filters.json
new file mode 100644
index 000000000..dac745b3a
--- /dev/null
+++ b/pipeline/clean/opuscleaner/configs/en-ja/default.filters.json
@@ -0,0 +1,77 @@
+{
+  "version": 1,
+  "files": [
+  ],
+  "filters": [
+    {
+      "filter": "remove_empty_lines",
+      "parameters": {},
+      "language": null
+    },
+    {
+      "filter": "normalize_whitespace",
+      "parameters": {
+        "COLLAPSE": true
+      },
+      "language": "<src>"
+    },
+    {
+      "filter": "deescape-special-chars",
+      "parameters": {
+        "LANG1": "other"
+      },
+      "language": "<src>"
+    },
+    {
+      "filter": "max_length",
+      "parameters": {
+        "MAXLENGTH": 150,
+        "MINLENGTH": 1
+      },
+      "language": null
+    },
+    {
+      "filter": "fix_wiki",
+      "parameters": {
+        "ALWAYS": false,
+        "FOOTNOTES": true,
+        "URLS": true,
+        "WIKILINKS": true,
+        "CODE": true,
+        "HEADINGS": true,
+        "REMOVEEMPTYLINES": true
+      },
+      "language": null
+    },
+    {
+      "filter": "alpha_ratio",
+      "parameters": {
+        "LANG1": "<src>",
+        "LANG2": "<trg>",
+        "SRCWORDRAT": 0.4,
+        "TRGWORDRAT": 0.0,
+        "SRCALPHARAT": 0.5,
+        "TRGALPHARAT": 0.0,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "num_mismatch",
+      "parameters": {
+        "RATIO": 1,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "fasttext_filter",
+      "parameters": {
+        "FASTTEXT_MODEL_TYPE": "large",
+        "LANG1": "<src>",
+        "LANG2": "<trg>"
+      },
+      "language": null
+    }
+  ]
+}
diff --git a/pipeline/clean/opuscleaner/configs/en-ko/default.filters.json b/pipeline/clean/opuscleaner/configs/en-ko/default.filters.json
new file mode 100644
index 000000000..dac745b3a
--- /dev/null
+++ b/pipeline/clean/opuscleaner/configs/en-ko/default.filters.json
@@ -0,0 +1,77 @@
+{
+  "version": 1,
+  "files": [
+  ],
+  "filters": [
+    {
+      "filter": "remove_empty_lines",
+      "parameters": {},
+      "language": null
+    },
+    {
+      "filter": "normalize_whitespace",
+      "parameters": {
+        "COLLAPSE": true
+      },
+      "language": "<src>"
+    },
+    {
+      "filter": "deescape-special-chars",
+      "parameters": {
+        "LANG1": "other"
+      },
+      "language": "<src>"
+    },
+    {
+      "filter": "max_length",
+      "parameters": {
+        "MAXLENGTH": 150,
+        "MINLENGTH": 1
+      },
+      "language": null
+    },
+    {
+      "filter": "fix_wiki",
+      "parameters": {
+        "ALWAYS": false,
+        "FOOTNOTES": true,
+        "URLS": true,
+        "WIKILINKS": true,
+        "CODE": true,
+        "HEADINGS": true,
+        "REMOVEEMPTYLINES": true
+      },
+      "language": null
+    },
+    {
+      "filter": "alpha_ratio",
+      "parameters": {
+        "LANG1": "<src>",
+        "LANG2": "<trg>",
+        "SRCWORDRAT": 0.4,
+        "TRGWORDRAT": 0.0,
+        "SRCALPHARAT": 0.5,
+        "TRGALPHARAT": 0.0,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "num_mismatch",
+      "parameters": {
+        "RATIO": 1,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "fasttext_filter",
+      "parameters": {
+        "FASTTEXT_MODEL_TYPE": "large",
+        "LANG1": "<src>",
+        "LANG2": "<trg>"
+      },
+      "language": null
+    }
+  ]
+}
diff --git a/pipeline/clean/opuscleaner/configs/en-zh/default.filters.json b/pipeline/clean/opuscleaner/configs/en-zh/default.filters.json
new file mode 100644
index 000000000..dac745b3a
--- /dev/null
+++ b/pipeline/clean/opuscleaner/configs/en-zh/default.filters.json
@@ -0,0 +1,77 @@
+{
+  "version": 1,
+  "files": [
+  ],
+  "filters": [
+    {
+      "filter": "remove_empty_lines",
+      "parameters": {},
+      "language": null
+    },
+    {
+      "filter": "normalize_whitespace",
+      "parameters": {
+        "COLLAPSE": true
+      },
+      "language": "<src>"
+    },
+    {
+      "filter": "deescape-special-chars",
+      "parameters": {
+        "LANG1": "other"
+      },
+      "language": "<src>"
+    },
+    {
+      "filter": "max_length",
+      "parameters": {
+        "MAXLENGTH": 150,
+        "MINLENGTH": 1
+      },
+      "language": null
+    },
+    {
+      "filter": "fix_wiki",
+      "parameters": {
+        "ALWAYS": false,
+        "FOOTNOTES": true,
+        "URLS": true,
+        "WIKILINKS": true,
+        "CODE": true,
+        "HEADINGS": true,
+        "REMOVEEMPTYLINES": true
+      },
+      "language": null
+    },
+    {
+      "filter": "alpha_ratio",
+      "parameters": {
+        "LANG1": "<src>",
+        "LANG2": "<trg>",
+        "SRCWORDRAT": 0.4,
+        "TRGWORDRAT": 0.0,
+        "SRCALPHARAT": 0.5,
+        "TRGALPHARAT": 0.0,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "num_mismatch",
+      "parameters": {
+        "RATIO": 1,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "fasttext_filter",
+      "parameters": {
+        "FASTTEXT_MODEL_TYPE": "large",
+        "LANG1": "<src>",
+        "LANG2": "<trg>"
+      },
+      "language": null
+    }
+  ]
+}
diff --git a/pipeline/clean/opuscleaner/configs/ja-en/default.filters.json b/pipeline/clean/opuscleaner/configs/ja-en/default.filters.json
new file mode 100644
index 000000000..dac86c6c4
--- /dev/null
+++ b/pipeline/clean/opuscleaner/configs/ja-en/default.filters.json
@@ -0,0 +1,77 @@
+{
+  "version": 1,
+  "files": [
+  ],
+  "filters": [
+    {
+      "filter": "remove_empty_lines",
+      "parameters": {},
+      "language": null
+    },
+    {
+      "filter": "normalize_whitespace",
+      "parameters": {
+        "COLLAPSE": true
+      },
+      "language": "<trg>"
+    },
+    {
+      "filter": "deescape-special-chars",
+      "parameters": {
+        "LANG1": "other"
+      },
+      "language": "<trg>"
+    },
+    {
+      "filter": "max_length",
+      "parameters": {
+        "MAXLENGTH": 150,
+        "MINLENGTH": 1
+      },
+      "language": null
+    },
+    {
+      "filter": "fix_wiki",
+      "parameters": {
+        "ALWAYS": false,
+        "FOOTNOTES": true,
+        "URLS": true,
+        "WIKILINKS": true,
+        "CODE": true,
+        "HEADINGS": true,
+        "REMOVEEMPTYLINES": true
+      },
+      "language": null
+    },
+    {
+      "filter": "alpha_ratio",
+      "parameters": {
+        "LANG1": "<src>",
+        "LANG2": "<trg>",
+        "SRCWORDRAT": 0.0,
+        "TRGWORDRAT": 0.4,
+        "SRCALPHARAT": 0.0,
+        "TRGALPHARAT": 0.5,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "num_mismatch",
+      "parameters": {
+        "RATIO": 1,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "fasttext_filter",
+      "parameters": {
+        "FASTTEXT_MODEL_TYPE": "large",
+        "LANG1": "<src>",
+        "LANG2": "<trg>"
+      },
+      "language": null
+    }
+  ]
+}
diff --git a/pipeline/clean/opuscleaner/configs/ko-en/default.filters.json b/pipeline/clean/opuscleaner/configs/ko-en/default.filters.json
new file mode 100644
index 000000000..dac86c6c4
--- /dev/null
+++ b/pipeline/clean/opuscleaner/configs/ko-en/default.filters.json
@@ -0,0 +1,77 @@
+{
+  "version": 1,
+  "files": [
+  ],
+  "filters": [
+    {
+      "filter": "remove_empty_lines",
+      "parameters": {},
+      "language": null
+    },
+    {
+      "filter": "normalize_whitespace",
+      "parameters": {
+        "COLLAPSE": true
+      },
+      "language": "<trg>"
+    },
+    {
+      "filter": "deescape-special-chars",
+      "parameters": {
+        "LANG1": "other"
+      },
+      "language": "<trg>"
+    },
+    {
+      "filter": "max_length",
+      "parameters": {
+        "MAXLENGTH": 150,
+        "MINLENGTH": 1
+      },
+      "language": null
+    },
+    {
+      "filter": "fix_wiki",
+      "parameters": {
+        "ALWAYS": false,
+        "FOOTNOTES": true,
+        "URLS": true,
+        "WIKILINKS": true,
+        "CODE": true,
+        "HEADINGS": true,
+        "REMOVEEMPTYLINES": true
+      },
+      "language": null
+    },
+    {
+      "filter": "alpha_ratio",
+      "parameters": {
+        "LANG1": "<src>",
+        "LANG2": "<trg>",
+        "SRCWORDRAT": 0.0,
+        "TRGWORDRAT": 0.4,
+        "SRCALPHARAT": 0.0,
+        "TRGALPHARAT": 0.5,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "num_mismatch",
+      "parameters": {
+        "RATIO": 1,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "fasttext_filter",
+      "parameters": {
+        "FASTTEXT_MODEL_TYPE": "large",
+        "LANG1": "<src>",
+        "LANG2": "<trg>"
+      },
+      "language": null
+    }
+  ]
+}
diff --git a/pipeline/clean/opuscleaner/configs/zh-en/default.filters.json b/pipeline/clean/opuscleaner/configs/zh-en/default.filters.json
new file mode 100644
index 000000000..dac86c6c4
--- /dev/null
+++ b/pipeline/clean/opuscleaner/configs/zh-en/default.filters.json
@@ -0,0 +1,77 @@
+{
+  "version": 1,
+  "files": [
+  ],
+  "filters": [
+    {
+      "filter": "remove_empty_lines",
+      "parameters": {},
+      "language": null
+    },
+    {
+      "filter": "normalize_whitespace",
+      "parameters": {
+        "COLLAPSE": true
+      },
+      "language": "<trg>"
+    },
+    {
+      "filter": "deescape-special-chars",
+      "parameters": {
+        "LANG1": "other"
+      },
+      "language": "<trg>"
+    },
+    {
+      "filter": "max_length",
+      "parameters": {
+        "MAXLENGTH": 150,
+        "MINLENGTH": 1
+      },
+      "language": null
+    },
+    {
+      "filter": "fix_wiki",
+      "parameters": {
+        "ALWAYS": false,
+        "FOOTNOTES": true,
+        "URLS": true,
+        "WIKILINKS": true,
+        "CODE": true,
+        "HEADINGS": true,
+        "REMOVEEMPTYLINES": true
+      },
+      "language": null
+    },
+    {
+      "filter": "alpha_ratio",
+      "parameters": {
+        "LANG1": "<src>",
+        "LANG2": "<trg>",
+        "SRCWORDRAT": 0.0,
+        "TRGWORDRAT": 0.4,
+        "SRCALPHARAT": 0.0,
+        "TRGALPHARAT": 0.5,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "num_mismatch",
+      "parameters": {
+        "RATIO": 1,
+        "DEBUG": false
+      },
+      "language": null
+    },
+    {
+      "filter": "fasttext_filter",
+      "parameters": {
+        "FASTTEXT_MODEL_TYPE": "large",
+        "LANG1": "<src>",
+        "LANG2": "<trg>"
+      },
+      "language": null
+    }
+  ]
+}
diff --git a/pipeline/clean/opuscleaner/generate_filters.py b/pipeline/clean/opuscleaner/generate_filters.py
index f30420650..f5316d1a2 100644
--- a/pipeline/clean/opuscleaner/generate_filters.py
+++ b/pipeline/clean/opuscleaner/generate_filters.py
@@ -33,8 +33,7 @@ def find_custom_filter(src: str, trg: str, dataset: str) -> Optional[str]:
     paths = [
         f"{CURRENT_FOLDER}/configs/{src}-{trg}/{dataset}.filters.json",
         f"{CURRENT_FOLDER}/configs/{src}-{trg}/{dataset_opus}.filters.json",
-        f"{CURRENT_FOLDER}/configs/{trg}-{src}/{dataset}.filters.json",
-        f"{CURRENT_FOLDER}/configs/{trg}-{src}/{dataset_opus}.filters.json",
+        f"{CURRENT_FOLDER}/configs/{src}-{trg}/default.filters.json",
         f"{CURRENT_FOLDER}/configs/{dataset}.filters.json",
         f"{CURRENT_FOLDER}/configs/{dataset_opus}.filters.json",
     ]
diff --git a/pipeline/clean/tools/clean_mono.py b/pipeline/clean/tools/clean_mono.py
index 0df09ec42..c3e7a01c8 100755
--- a/pipeline/clean/tools/clean_mono.py
+++ b/pipeline/clean/tools/clean_mono.py
@@ -31,7 +31,11 @@ def main():
 
 
 def clean_mono(src, lang):
-    src_toks = src.split()
+    # TODO: move mono cleaning to OpusCleaner
+    #  when it support this https://github.com/hplt-project/OpusCleaner/issues/141
+
+    # treat individual characters as tokens for CJK
+    src_toks = src.split() if lang not in {"zh", "ja", "ko"} else src
     src_len = len(src_toks)
 
     if not src_len:
diff --git a/taskcluster/kinds/clean-corpus/kind.yml b/taskcluster/kinds/clean-corpus/kind.yml
index b212cd668..04c59230d 100644
--- a/taskcluster/kinds/clean-corpus/kind.yml
+++ b/taskcluster/kinds/clean-corpus/kind.yml
@@ -38,6 +38,12 @@ tasks:
                     - pipeline/clean/opuscleaner/configs/remove_frequent_patterns.txt
                     - pipeline/clean/opuscleaner/configs/default.filters.json
                     - pipeline/clean/opuscleaner/configs/ru-en/opus_ELRC-3075-wikipedia_health-v1.filters.json
+                    - pipeline/clean/opuscleaner/configs/en-zh/default.filters.json
+                    - pipeline/clean/opuscleaner/configs/zh-en/default.filters.json
+                    - pipeline/clean/opuscleaner/configs/en-ja/default.filters.json
+                    - pipeline/clean/opuscleaner/configs/ja-en/default.filters.json
+                    - pipeline/clean/opuscleaner/configs/en-ko/default.filters.json
+                    - pipeline/clean/opuscleaner/configs/ko-en/default.filters.json
                     - pipeline/clean/opuscleaner/configs/opus_LinguaTools-WikiTitles-v2014.filters.json
                     - pipeline/clean/opuscleaner/configs/opus_NLLB-v1.filters.json
                     - pipeline/clean/opuscleaner/configs/opus_OpenSubtitles-v2018.filters.json
diff --git a/tests/test_filters.py b/tests/test_filters.py
index e07610a60..c85621e56 100644
--- a/tests/test_filters.py
+++ b/tests/test_filters.py
@@ -28,12 +28,13 @@ def data_dir():
             Mode.custom,
             "ru-en/opus_ELRC-3075-wikipedia_health-v1.filters.json",
         ),
+        # backward direction should have a separate custom config because it can be different (for example, zh)
         (
             "en",
             "ru",
             "opus_ELRC-3075-wikipedia_health/v1",
             Mode.custom,
-            "ru-en/opus_ELRC-3075-wikipedia_health-v1.filters.json",
+            "default.filters.json",
         ),
         # verify dataset specific config is used for different language pairs
         ("ru", "en", "opus_UNPC/v1.0", Mode.custom, "opus_UNPC-v1.0.filters.json"),
@@ -47,6 +48,21 @@ def data_dir():
             "default.filters.json",
         ),
         ("fr", "en", "opus_UNPC/v1.0", Mode.defaults, "default.filters.json"),
+        # make sure Chinese uses language level default configs
+        (
+            "zh",
+            "en",
+            "opus_UNPC/v1.0",
+            Mode.custom,
+            "zh-en/default.filters.json",
+        ),
+        (
+            "en",
+            "zh",
+            "opus_UNPC/v1.0",
+            Mode.custom,
+            "en-zh/default.filters.json",
+        ),
     ],
     ids=[
         "default-en-ru",
@@ -58,6 +74,8 @@ def data_dir():
         "dataset-fr-en",
         "override-with-default-ru-en-elrc",
         "override-with-default-fr-en-unpc",
+        "zh-en",
+        "en-zh",
     ],
 )
 def test_generate_filters(params, data_dir):
@@ -79,13 +97,27 @@ def test_generate_filters(params, data_dir):
             expected = json.load(f_conf)
     assert len(actual["filters"]) == len(expected["filters"])
     assert {f["filter"] for f in actual["filters"]} == {f["filter"] for f in expected["filters"]}
-    assert {f["language"] for f in actual["filters"] if f["filter"] == "normalize_whitespace"} == {
-        src,
-        trg,
+    # check languages in whitespace filters where there are two of them
+    whitespace_filters = {
+        f["language"] for f in actual["filters"] if f["filter"] == "normalize_whitespace"
     }
+    if len(whitespace_filters) == 2:
+        assert whitespace_filters == {
+            src,
+            trg,
+        }
     # max length value is slightly changed in opus_ELRC-3075-wikipedia_health/v1 to verify that this is the same config
     assert [f for f in actual["filters"] if f["filter"] == "max_length"][0]["parameters"][
         "MAXLENGTH"
     ] == [f for f in expected["filters"] if f["filter"] == "max_length"][0]["parameters"][
         "MAXLENGTH"
     ]
+    # alpha ratios are different for zh
+    alpha_ratio_filters = [f for f in actual["filters"] if f["filter"] == "alpha_ratio"]
+    if alpha_ratio_filters:
+        assert (
+            alpha_ratio_filters[0]["parameters"]["SRCWORDRAT"]
+            == [f for f in expected["filters"] if f["filter"] == "alpha_ratio"][0]["parameters"][
+                "SRCWORDRAT"
+            ]
+        )
diff --git a/utils/config_generator.py b/utils/config_generator.py
index c14c03a8c..aa6d5c181 100644
--- a/utils/config_generator.py
+++ b/utils/config_generator.py
@@ -118,6 +118,9 @@ def update_config(
     else:
         experiment["pretrained-models"] = {}
 
+    if source in CJK_LANGS or target in CJK_LANGS:
+        experiment["opuscleaner-mode"] = "custom"
+
     datasets = prod_config["datasets"]
 
     # Clear out the base config.