diff --git a/.github/workflows/llm_performance_tests.yml b/.github/workflows/llm_performance_tests.yml
index 292bd52c..59849ba4 100644
--- a/.github/workflows/llm_performance_tests.yml
+++ b/.github/workflows/llm_performance_tests.yml
@@ -772,6 +772,97 @@ jobs:
 
           call conda deactivate
 
+      - name: Prepare igpu perf test (int4+fp16 1024-128)
+        shell: bash
+        run: |
+          sed -i 's/{today}_test3/{today}_test1/g' python/llm/dev/benchmark/all-in-one/run.py
+          sed -i "s/path to your local model hub/$MODEL_HUB_DIR/g" python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16.yaml
+
+      - name: Test on igpu (int4+fp16 1024-128)
+        shell: cmd
+        run: |
+          call conda activate igpu-perf
+          pip install transformers==4.31.0
+
+          call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
+          set SYCL_CACHE_PERSISTENT=1
+          set BIGDL_LLM_XMX_DISABLED=1
+          REM for llava
+          set TRANSFORMERS_OFFLINE=1
+
+          cd python\llm\dev\benchmark\all-in-one
+          move ..\..\..\test\benchmark\igpu-perf\1024-128_int4_fp16.yaml config.yaml
+          set PYTHONIOENCODING=utf-8
+          python run.py >> %CSV_SAVE_PATH%\1024-128_int4_fp16\log\%LOG_FILE% 2>&1
+          if %ERRORLEVEL% neq 0 (exit /b 1)
+
+          call conda deactivate
+
+      - name: Prepare igpu perf test for Mistral (int4+fp16 1024-128)
+        shell: bash
+        run: |
+          sed -i 's/{today}_test1/{today}_test2/g' python/llm/dev/benchmark/all-in-one/run.py
+          sed -i "s/path to your local model hub/$MODEL_HUB_DIR/g" python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_434.yaml
+
+      - name: Test on igpu for Mistral (int4+fp16 1024-128)
+        shell: cmd
+        run: |
+          call conda activate igpu-perf
+          pip install transformers==4.34.0
+
+          call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
+          set SYCL_CACHE_PERSISTENT=1
+          set BIGDL_LLM_XMX_DISABLED=1
+
+          cd python\llm\dev\benchmark\all-in-one
+          move ..\..\..\test\benchmark\igpu-perf\1024-128_int4_fp16_434.yaml config.yaml
+          set PYTHONIOENCODING=utf-8
+          python run.py >> %CSV_SAVE_PATH%\1024-128_int4_fp16\log\%LOG_FILE% 2>&1
+          if %ERRORLEVEL% neq 0 (exit /b 1)
+
+          call conda deactivate
+
+      - name: Prepare igpu perf test for Qwen 1.5 (int4+fp16 1024-128)
+        shell: bash
+        run: |
+          sed -i 's/{today}_test2/{today}_test3/g' python/llm/dev/benchmark/all-in-one/run.py
+          sed -i "s/path to your local model hub/$MODEL_HUB_DIR/g" python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_437.yaml
+
+      - name: Test on igpu for Qwen 1.5 (int4+fp16 1024-128)
+        shell: cmd
+        run: |
+          call conda activate igpu-perf
+          pip install transformers==4.37.0
+
+          call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
+          set SYCL_CACHE_PERSISTENT=1
+          set BIGDL_LLM_XMX_DISABLED=1
+
+          cd python\llm\dev\benchmark\all-in-one
+          move ..\..\..\test\benchmark\igpu-perf\1024-128_int4_fp16_437.yaml config.yaml
+          set PYTHONIOENCODING=utf-8
+          python run.py >> %CSV_SAVE_PATH%\1024-128_int4_fp16\log\%LOG_FILE% 2>&1
+          if %ERRORLEVEL% neq 0 (exit /b 1)
+
+          call conda deactivate
+
+      - name: Concat csv and generate html (int4+fp16 1024-128)
+        shell: cmd
+        run: |
+          call conda activate html-gen
+
+          cd python\llm\dev\benchmark\all-in-one
+          python ..\..\..\test\benchmark\concat_csv.py
+          if %ERRORLEVEL% neq 0 (exit /b 1)
+          del /q *test*.csv
+          move *.csv %CSV_SAVE_PATH%\1024-128_int4_fp16\
+          cd ..\..\..\test\benchmark
+          python csv_to_html.py -f %CSV_SAVE_PATH%\1024-128_int4_fp16\
+          if %ERRORLEVEL% neq 0 (exit /b 1)
+          move %CSV_SAVE_PATH%\1024-128_int4_fp16\*.html %CSV_SAVE_PATH%
+
+          call conda deactivate
+
       - name: Upload results to ftp
         if: ${{ always() }}
         shell: cmd
diff --git a/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16.yaml b/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16.yaml
new file mode 100644
index 00000000..d4142ce8
--- /dev/null
+++ b/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16.yaml
@@ -0,0 +1,29 @@
+repo_id:
+  - 'THUDM/chatglm2-6b'
+  - 'THUDM/chatglm3-6b'
+  - 'baichuan-inc/Baichuan2-7B-Chat'
+  - 'baichuan-inc/Baichuan2-13B-Chat'
+  - 'internlm/internlm-chat-7b-8k'
+  - 'Qwen/Qwen-7B-Chat'
+  - 'BAAI/AquilaChat2-7B'
+  - '01-ai/Yi-6B'
+  - 'meta-llama/Llama-2-7b-chat-hf'
+  - 'meta-llama/Llama-2-13b-chat-hf'
+  - 'WisdomShell/CodeShell-7B-Chat'
+  - 'tiiuae/falcon-7b-instruct-with-patch'
+  - 'mosaicml/mpt-7b-chat'
+  - 'liuhaotian/llava-v1.5-7b'
+  # - 'RWKV/rwkv-4-world-7b'
+  # - 'RWKV/rwkv-5-world-7b'
+  - 'IEITYuan/Yuan2-2B-hf'
+local_model_hub: 'path to your local model hub'
+warm_up: 1
+num_trials: 3
+num_beams: 1 # default to greedy search
+low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
+batch_size: 1 # default to 1
+in_out_pairs:
+  - '1024-128'
+test_api:
+  - "transformer_int4_fp16_gpu_win" # on Intel GPU for Windows, use fp16 for non-linear layer
+cpu_embedding: True # whether put embedding to CPU (only avaiable now for gpu win related test_api)
diff --git a/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_434.yaml b/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_434.yaml
new file mode 100644
index 00000000..57f0a3d3
--- /dev/null
+++ b/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_434.yaml
@@ -0,0 +1,13 @@
+repo_id:
+  - 'mistralai/Mistral-7B-Instruct-v0.1'
+local_model_hub: 'path to your local model hub'
+warm_up: 1
+num_trials: 3
+num_beams: 1 # default to greedy search
+low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
+batch_size: 1 # default to 1
+in_out_pairs:
+  - '1024-128'
+test_api:
+  - "transformer_int4_fp16_gpu_win" # on Intel GPU for Windows, use fp16 for non-linear layer
+cpu_embedding: True # whether put embedding to CPU (only avaiable now for gpu win related test_api)
diff --git a/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_437.yaml b/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_437.yaml
new file mode 100644
index 00000000..2e2ec3c5
--- /dev/null
+++ b/python/llm/test/benchmark/igpu-perf/1024-128_int4_fp16_437.yaml
@@ -0,0 +1,13 @@
+repo_id:
+  - 'Qwen/Qwen1.5-7B-Chat'
+local_model_hub: 'path to your local model hub'
+warm_up: 1
+num_trials: 3
+num_beams: 1 # default to greedy search
+low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
+batch_size: 1 # default to 1
+in_out_pairs:
+  - '1024-128'
+test_api:
+  - "transformer_int4_fp16_gpu_win" # on Intel GPU for Windows, use fp16 for non-linear layer
+cpu_embedding: True # whether put embedding to CPU (only avaiable now for gpu win related test_api)