diff --git a/docs/install/Tables.md b/docs/install/Tables.md
index 450cfac852a..9809966b93f 100644
--- a/docs/install/Tables.md
+++ b/docs/install/Tables.md
@@ -468,66 +468,68 @@ PaddePaddle 通过编译时指定路径来实现引用各种 BLAS/CUDA/cuDNN 库
     </thead>
     <tbody>
     <tr>
-    <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm2.5-volta </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm2.5-ampere </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <tr>
-    <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm2.5-volta</td>
-    <td> - </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-linux_x86_64.whl"></a>paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-linux_x86_64.whl</td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm2.5-ampere</td>
-    <td> - </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl"></a>paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl</td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm2.5-volta </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-win_amd64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm2.5-ampere </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-win_amd64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm2.5-volta </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-win_amd64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm2.5-ampere </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-win_amd64.whl</a></td>
+        <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm-volta </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp37-cp37m-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm-ampere </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp37-cp37m-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm-volta</td>
+        <td> - </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm-ampere</td>
+        <td> - </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm-volta </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp311-cp311-win_amd64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm-ampere </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp311-cp311-win_amd64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm-volta </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp311-cp311-win_amd64.wh</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm-ampere </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp311-cp311-win_amd64.whl</a> </td>
     </tr>
    </tbody>
 </table>
diff --git a/docs/install/Tables_en.md b/docs/install/Tables_en.md
index adeffb0206a..5398adeece0 100644
--- a/docs/install/Tables_en.md
+++ b/docs/install/Tables_en.md
@@ -452,7 +452,7 @@ Please note that: in the commands, <code> paddlepaddle-gpu==2.2.1 </code> will i
 <table>
     <thead>
     <tr>
-        <th> 版本说明 </th>
+        <th> Release Instruction </th>
         <th> cp37-cp37m    </th>
         <th> cp38-cp38    </th>
         <th> cp39-cp39    </th>
@@ -462,66 +462,68 @@ Please note that: in the commands, <code> paddlepaddle-gpu==2.2.1 </code> will i
     </thead>
     <tbody>
     <tr>
-    <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm2.5-volta </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm2.5-ampere </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <tr>
-    <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm2.5-volta</td>
-    <td> - </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-linux_x86_64.whl"></a>paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-linux_x86_64.whl</td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm2.5-ampere</td>
-    <td> - </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl"></a>paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-linux_x86_64.whl</td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-linux_x86_64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-linux_x86_64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm2.5-volta </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+volta.llm-cp311-cp311-win_amd64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm2.5-ampere </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post118.dev0+ampere.llm-cp311-cp311-win_amd64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm2.5-volta </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+volta.llm-cp311-cp311-win_amd64.whl</a></td>
-    </tr>
-    <tr>
-    <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm2.5-ampere </td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp37-cp37m-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp38-cp38-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp39-cp39-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp310-cp310-win_amd64.whl</a></td>
-    <td><a herf="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/4b5a9537218232f8679913ed068762b5d227ac6d/paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.5.2.post120.dev0+ampere.llm-cp311-cp311-win_amd64.whl</a></td>
+        <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm-volta </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp37-cp37m-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> linux-cuda11.8-cudnn8.6-trt8.5-mkl-gcc8.2-avx-llm-ampere </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp37-cp37m-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp37-cp37m-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Gcc8.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm-volta</td>
+        <td> - </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td>linux-cuda12.0-cudnn8.9-trt8.6-mkl-gcc12.2-avx-llm-ampere</td>
+        <td> - </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp38-cp38-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp38-cp38-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp39-cp39-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp39-cp39-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp310-cp310-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp310-cp310-linux_x86_64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Linux-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Gcc12.2/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp311-cp311-linux_x86_64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp311-cp311-linux_x86_64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm-volta </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.volta.llm-cp311-cp311-win_amd64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda11.8-cudnn8.6-trt8.5-mkl-vs2019-avx-llm-ampere </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda11.8-Cudnn8.6-Trt8.5-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu118.ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu118.ampere.llm-cp311-cp311-win_amd64.whl</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm-volta </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.volta.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.volta.llm-cp311-cp311-win_amd64.wh</a> </td>
+    </tr>
+    <tr>
+        <td> windows-cuda12.0-cudnn8.9-trt8.6-mkl-vs2019-avx-llm-ampere </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp37-cp37m-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp37-cp37m-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp38-cp38-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp38-cp38-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp39-cp39-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp39-cp39-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp310-cp310-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp310-cp310-win_amd64.whl</a> </td>
+        <td> <a href="https://paddle-qa.bj.bcebos.com/paddle-pipeline/LLM-Windows-Gpu-Cuda12.0-Cudnn8.9-Trt8.6-Mkl-Avx-Vs2019/latest/paddlepaddle_gpu-2.6.0.dev0%2Bcu120.ampere.llm-cp311-cp311-win_amd64.whl">paddlepaddle_gpu-2.6.0.dev0+cu120.ampere.llm-cp311-cp311-win_amd64.whl</a> </td>
     </tr>
    </tbody>
 </table>