diff --git a/docs/ascend_tutorial/ascend_quick_start.rst b/docs/ascend_tutorial/ascend_quick_start.rst
index 390c864d899..90bf0aebaab 100644
--- a/docs/ascend_tutorial/ascend_quick_start.rst
+++ b/docs/ascend_tutorial/ascend_quick_start.rst
@@ -187,6 +187,8 @@ vllm & vllm-ascend
 +-----------+-------------------------+-------------+-------------------+-------------------+-------------------+--------------------------+
 |   DAPO    | Qwen3-14B-base          |    5.9%     |        pending    |        FSDP       |    vllm-ascend    |    Atlas 200T A2 Box16   |
 +-----------+-------------------------+-------------+-------------------+-------------------+-------------------+--------------------------+
+|   DAPO    | Qwen3-30B-base          |    1.08%    |        pending    |        FSDP       |    vllm-ascend    |    Atlas 200T A2 Box16   |
++-----------+-------------------------+-------------+-------------------+-------------------+-------------------+--------------------------+
 
 **表2** SFT类算法
 
diff --git a/recipe/dapo/run_dapo_qwen3_moe_30b_base_npu_fsdp.sh b/recipe/dapo/run_dapo_qwen3_moe_30b_base_npu_fsdp.sh
new file mode 100644
index 00000000000..36cf175a18f
--- /dev/null
+++ b/recipe/dapo/run_dapo_qwen3_moe_30b_base_npu_fsdp.sh
@@ -0,0 +1,146 @@
+#!/usr/bin/env bash
+set -euxo pipefail
+
+project_name='DAPO'
+exp_name='DAPO-Qwen3-MOE-30B-FSDP-128rank-gbs512'
+
+NNODES=8
+NPUS_PER_NODE=16
+
+adv_estimator=grpo
+
+use_kl_in_reward=False
+kl_coef=0.0
+use_kl_loss=False
+kl_loss_coef=0.0
+
+clip_ratio_low=0.2
+clip_ratio_high=0.28
+
+max_prompt_length=$((1024 * 2))
+max_response_length=$((1024 * 20))
+enable_overlong_buffer=True
+overlong_buffer_len=$((1024 * 4))
+overlong_penalty_factor=1.0
+loss_agg_mode="token-mean"
+ppo_mini_batch_size=32
+
+enable_filter_groups=True
+filter_groups_metric=acc
+max_num_gen_batches=10
+train_prompt_bsz=512
+gen_prompt_bsz=$((train_prompt_bsz * 3))
+n_resp_per_prompt=16
+
+RAY_ADDRESS=${RAY_ADDRESS:-"http://localhost:8265"}
+WORKING_DIR=${WORKING_DIR:-"${PWD}"}
+RUNTIME_ENV=${RUNTIME_ENV:-"${WORKING_DIR}/verl/trainer/runtime_env.yaml"}
+
+# Paths
+RAY_DATA_HOME=${RAY_DATA_HOME:-"${HOME}/verl"}
+MODEL_PATH=${MODEL_PATH:-"${RAY_DATA_HOME}/models/Qwen3-30B-A3B-Base"}
+CKPTS_DIR=${CKPTS_DIR:-"${RAY_DATA_HOME}/ckpts/${project_name}/${exp_name}"}
+TRAIN_FILE=${TRAIN_FILE:-"${RAY_DATA_HOME}/data/dapo-math-17k.parquet"}
+TEST_FILE=${TEST_FILE:-"${RAY_DATA_HOME}/data/aime-2024.parquet"}
+
+# Algorithm
+temperature=1.0
+top_p=1.0
+top_k=-1 # 0 for HF rollout, -1 for vLLM rollout
+val_top_p=0.7
+
+# Performance Related Parameter
+sp_size=16 # For load-balance. For smaller cluster this can be set to as less as 2.
+use_dynamic_bsz=True
+actor_ppo_max_token_len=$(((max_prompt_length + max_response_length) / 2))
+infer_ppo_max_token_len=$(((max_prompt_length + max_response_length) / 2))
+offload=True
+recompute=True
+max_num_seqs=128
+gen_tp=2
+gen_world_size=$((NNODES * NPUS_PER_NODE)) # nnodes* npus_in_per_node
+
+
+ray job submit --no-wait --runtime-env="${RUNTIME_ENV}" \
+    -- python3 -m recipe.dapo.main_dapo \
+    data.train_files="${TRAIN_FILE}" \
+    data.val_files="${TEST_FILE}" \
+    data.prompt_key=prompt \
+    data.truncation='left' \
+    data.max_prompt_length=${max_prompt_length} \
+    data.max_response_length=${max_response_length} \
+    data.gen_batch_size=${gen_prompt_bsz} \
+    data.train_batch_size=${train_prompt_bsz} \
+    actor_rollout_ref.rollout.name=vllm \
+    actor_rollout_ref.rollout.n=${n_resp_per_prompt} \
+    actor_rollout_ref.rollout.max_num_seqs=${max_num_seqs} \
+    actor_rollout_ref.rollout.max_num_batched_tokens=$((max_prompt_length + max_response_length))  \
+    algorithm.adv_estimator=${adv_estimator} \
+    algorithm.use_kl_in_reward=${use_kl_in_reward} \
+    algorithm.kl_ctrl.kl_coef=${kl_coef} \
+    actor_rollout_ref.actor.use_kl_loss=${use_kl_loss} \
+    actor_rollout_ref.actor.kl_loss_coef=${kl_loss_coef} \
+    actor_rollout_ref.actor.clip_ratio_low=${clip_ratio_low} \
+    actor_rollout_ref.actor.clip_ratio_high=${clip_ratio_high} \
+    actor_rollout_ref.actor.clip_ratio_c=10.0 \
+    algorithm.filter_groups.enable=${enable_filter_groups} \
+    algorithm.filter_groups.max_num_gen_batches=${max_num_gen_batches} \
+    algorithm.filter_groups.metric=${filter_groups_metric} \
+    actor_rollout_ref.model.use_remove_padding=True \
+    actor_rollout_ref.actor.use_dynamic_bsz=${use_dynamic_bsz} \
+    actor_rollout_ref.ref.log_prob_use_dynamic_bsz=${use_dynamic_bsz} \
+    actor_rollout_ref.rollout.log_prob_use_dynamic_bsz=${use_dynamic_bsz} \
+    actor_rollout_ref.actor.ppo_max_token_len_per_gpu=${actor_ppo_max_token_len} \
+    actor_rollout_ref.ref.log_prob_max_token_len_per_gpu=${infer_ppo_max_token_len} \
+    actor_rollout_ref.rollout.log_prob_max_token_len_per_gpu=${infer_ppo_max_token_len} \
+    actor_rollout_ref.model.path="${MODEL_PATH}" \
+    +actor_rollout_ref.model.override_config.attention_dropout=0. \
+    +actor_rollout_ref.model.override_config.embd_pdrop=0. \
+    +actor_rollout_ref.model.override_config.resid_pdrop=0. \
+    actor_rollout_ref.model.enable_gradient_checkpointing=${recompute} \
+    actor_rollout_ref.actor.optim.lr=1e-6 \
+    actor_rollout_ref.actor.optim.lr_warmup_steps=10 \
+    actor_rollout_ref.actor.optim.weight_decay=0.1 \
+    actor_rollout_ref.actor.ppo_mini_batch_size=${ppo_mini_batch_size} \
+    actor_rollout_ref.actor.fsdp_config.param_offload=${offload} \
+    actor_rollout_ref.actor.fsdp_config.optimizer_offload=${offload} \
+    actor_rollout_ref.actor.fsdp_config.forward_prefetch=False \
+    actor_rollout_ref.actor.entropy_coeff=0 \
+    actor_rollout_ref.actor.grad_clip=1.0 \
+    actor_rollout_ref.actor.loss_agg_mode=${loss_agg_mode} \
+    actor_rollout_ref.actor.ulysses_sequence_parallel_size=${sp_size} \
+    actor_rollout_ref.rollout.gpu_memory_utilization=0.8 \
+    actor_rollout_ref.rollout.tensor_model_parallel_size=${gen_tp} \
+    +actor_rollout_ref.rollout.rollout_world_size=${gen_world_size} \
+    actor_rollout_ref.rollout.enable_chunked_prefill=True \
+    actor_rollout_ref.rollout.temperature=${temperature} \
+    actor_rollout_ref.rollout.top_p=${top_p} \
+    actor_rollout_ref.rollout.top_k=${top_k} \
+    actor_rollout_ref.rollout.val_kwargs.temperature=${temperature} \
+    actor_rollout_ref.rollout.val_kwargs.top_p=${val_top_p} \
+    actor_rollout_ref.rollout.val_kwargs.top_k=${top_k} \
+    actor_rollout_ref.rollout.val_kwargs.do_sample=True \
+    actor_rollout_ref.rollout.val_kwargs.n=1 \
+    actor_rollout_ref.ref.fsdp_config.param_offload=${offload} \
+    actor_rollout_ref.ref.ulysses_sequence_parallel_size=${sp_size} \
+    actor_rollout_ref.actor.fsdp_config.fsdp_size=-1 \
+    actor_rollout_ref.ref.fsdp_config.forward_prefetch=False \
+    actor_rollout_ref.rollout.enforce_eager=False \
+    actor_rollout_ref.rollout.free_cache_engine=True \
+    reward_model.reward_manager=dapo \
+    reward_model.overlong_buffer.enable=${enable_overlong_buffer} \
+    reward_model.overlong_buffer.len=${overlong_buffer_len} \
+    reward_model.overlong_buffer.penalty_factor=${overlong_penalty_factor} \
+    trainer.logger=['console','wandb'] \
+    trainer.project_name="${project_name}" \
+    trainer.experiment_name="${exp_name}" \
+    trainer.n_gpus_per_node="${NPUS_PER_NODE}" \
+    trainer.nnodes="${NNODES}" \
+    trainer.val_before_train=False \
+    trainer.test_freq=5 \
+    trainer.save_freq=-1 \
+    trainer.total_epochs=1 \
+    trainer.device="npu" \
+    actor_rollout_ref.actor.use_torch_compile=False \
+    actor_rollout_ref.ref.use_torch_compile=False 
+