diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index a122289c7..e22c7015f 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -20,6 +20,16 @@ jobs:
     - name: checkout code
       uses: actions/checkout@v4
 
+    - name: Set up Python
+      uses: actions/setup-python@v5
+      with:
+        python-version: '3.10'
+
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install colorama
+        
     - name: Check Format
       run: |
         pip install black
@@ -37,6 +47,33 @@ jobs:
       run: |
         pip install numpy
         pip install torch
+        pip install gguf
 
     - name: Python Test
       run: python scripts/python_test.py --cpu
+
+    - name: GGUF Generation Test
+      run: python scripts/gguf_test.py --mode gguf
+
+    - name: C++ Test on Linux
+      if: matrix.os == 'ubuntu-latest'
+      run: |
+        export LD_LIBRARY_PATH="$PWD/build/linux/x86_64/release:$LD_LIBRARY_PATH"
+        tests=(add gemm mul random_sample swiglu clip)
+        for t in "${tests[@]}"; do
+          build/linux/x86_64/release/infiniop-test \
+            test/infiniop-test/${t}.gguf \
+            --cpu \
+            --warmup 0 \
+            --run 0
+        done
+
+    - name: C++ Test on Windows
+      if: matrix.os == 'windows-latest'
+      run: |
+        $exe = ".\build\windows\x64\release\infiniop-test.exe"
+        $base = "test\infiniop-test"
+        $tests = @("add","gemm","mul","random_sample","swiglu","clip")
+        foreach ($t in $tests) {
+          & $exe "$base\$t.gguf" "--cpu" "--warmup" "0" "--run" "0"
+        }
diff --git a/scripts/gguf_test.py b/scripts/gguf_test.py
new file mode 100644
index 000000000..c882e492e
--- /dev/null
+++ b/scripts/gguf_test.py
@@ -0,0 +1,43 @@
+import os
+import sys
+import subprocess
+from set_env import set_env
+import argparse
+
+
+PROJECT_DIR = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
+os.chdir(PROJECT_DIR)
+
+def run(test, device, mode):
+    base = os.path.join(PROJECT_DIR, "test", "infiniop-test")
+    if mode in ("gguf", "all"):
+        if subprocess.run(f"python -m test_generate.testcases.{test}",
+                          shell=True, cwd=base).returncode:
+            return f"{test}: gguf failed"
+    if mode in ("cpp", "all"):
+        exe = os.path.join(PROJECT_DIR, "build/linux/x86_64/release/infiniop-test")
+        gguf = os.path.join(base, f"{test}.gguf")
+        if subprocess.run(f"{exe} {gguf} --{device} --warmup 20 --run 1000",
+                          shell=True, cwd=PROJECT_DIR).returncode:
+            return f"{test}: cpp failed"
+    return None
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--device", nargs="?", default="cpu")
+    parser.add_argument("--mode", choices=["gguf", "cpp", "all"], default="all")
+    args = parser.parse_args()
+
+    set_env()
+    tests = ["add", "gemm", "mul", "random_sample", "swiglu", "clip"]
+    fails = [r for t in tests if (r := run(t, args.device, args.mode))]
+    
+    if not fails:
+        print("\n\033[92mAll tests passed!\033[0m")
+        sys.exit(0)
+    print("\033[91mSome tests failed:\033[0m")
+    for f in fails: print(" -", f)
+    sys.exit(1)
+
+if __name__ == "__main__":
+    main()
diff --git a/test/infiniop-test/test_generate/testcases/add.py b/test/infiniop-test/test_generate/testcases/add.py
index b04ba2042..e94dc7b67 100644
--- a/test/infiniop-test/test_generate/testcases/add.py
+++ b/test/infiniop-test/test_generate/testcases/add.py
@@ -1,8 +1,6 @@
-from ast import List
 import numpy as np
 import gguf
 from typing import List
-from numpy.lib.stride_tricks import as_strided
 
 from .. import InfiniopTestWriter, InfiniopTestCase, np_dtype_to_ggml, gguf_strides, contiguous_gguf_strides, process_zero_stride_tensor
 
diff --git a/test/infiniop-test/test_generate/testcases/gemm.py b/test/infiniop-test/test_generate/testcases/gemm.py
index fb81c5522..a62986837 100644
--- a/test/infiniop-test/test_generate/testcases/gemm.py
+++ b/test/infiniop-test/test_generate/testcases/gemm.py
@@ -1,4 +1,3 @@
-from ast import List
 import numpy as np
 import gguf
 from typing import List
diff --git a/test/infiniop-test/test_generate/testcases/random_sample.py b/test/infiniop-test/test_generate/testcases/random_sample.py
index 15c07f2cc..559428fdb 100644
--- a/test/infiniop-test/test_generate/testcases/random_sample.py
+++ b/test/infiniop-test/test_generate/testcases/random_sample.py
@@ -1,7 +1,4 @@
-from ast import List
 import numpy as np
-import gguf
-from typing import List
 
 from .. import InfiniopTestWriter, InfiniopTestCase, np_dtype_to_ggml, gguf_strides