diff --git a/README.md b/README.md
index 7ebada73d..d1449abfc 100644
--- a/README.md
+++ b/README.md
@@ -259,8 +259,8 @@ python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct
 - ChatGLM
 - InternLM 2
 - Exaone 3
+- BaiChuan2
 - MiniCPM / MiniCPM 3
-
 **Embedding Models**
 
 - e5-mistral
diff --git a/python/sglang/srt/models/baichuan.py b/python/sglang/srt/models/baichuan.py
new file mode 100644
index 000000000..d699064b3
--- /dev/null
+++ b/python/sglang/srt/models/baichuan.py
@@ -0,0 +1,421 @@
+# coding=utf-8
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Inference-only BaiChuan model compatible with HuggingFace weights."""
+import math
+from typing import Iterable, Optional, Tuple
+
+import torch
+from torch import nn
+from transformers import PretrainedConfig
+from vllm.config import CacheConfig
+from vllm.distributed import (
+    get_tensor_model_parallel_rank,
+    get_tensor_model_parallel_world_size,
+)
+from vllm.model_executor.layers.linear import (
+    MergedColumnParallelLinear,
+    QKVParallelLinear,
+    RowParallelLinear,
+)
+from vllm.model_executor.layers.quantization.base_config import QuantizationConfig
+from vllm.model_executor.layers.rotary_embedding import get_rope
+from vllm.model_executor.layers.vocab_parallel_embedding import (
+    ParallelLMHead,
+    VocabParallelEmbedding,
+)
+from vllm.model_executor.model_loader.weight_utils import default_weight_loader
+
+from sglang.srt.layers.activation import SiluAndMul
+from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.logits_processor import LogitsProcessor
+from sglang.srt.layers.radix_attention import RadixAttention
+from sglang.srt.layers.sampler import Sampler
+from sglang.srt.model_executor.forward_batch_info import InputMetadata
+
+
+def _get_alibi_slopes(total_num_heads: int) -> torch.Tensor:
+    closest_power_of_2 = 2 ** math.floor(math.log2(total_num_heads))
+    base = torch.tensor(
+        2 ** (-(2 ** -(math.log2(closest_power_of_2) - 3))),
+        dtype=torch.float32,
+    )
+    powers = torch.arange(1, 1 + closest_power_of_2, dtype=torch.int32)
+    slopes = torch.pow(base, powers)
+
+    if closest_power_of_2 != total_num_heads:
+        extra_base = torch.tensor(
+            2 ** (-(2 ** -(math.log2(2 * closest_power_of_2) - 3))),
+            dtype=torch.float32,
+        )
+        num_remaining_heads = min(
+            closest_power_of_2, total_num_heads - closest_power_of_2
+        )
+        extra_powers = torch.arange(
+            start=1, end=1 + 2 * num_remaining_heads, step=2, dtype=torch.int32
+        )
+        slopes = torch.cat([slopes, torch.pow(extra_base, extra_powers)], dim=0)
+    return slopes
+
+
+class BaiChuanMLP(nn.Module):
+
+    def __init__(
+        self,
+        hidden_size: int,
+        intermediate_size: int,
+        hidden_act: str,
+        quant_config: Optional[QuantizationConfig] = None,
+    ):
+        super().__init__()
+        self.gate_up_proj = MergedColumnParallelLinear(
+            hidden_size, [intermediate_size] * 2, bias=False, quant_config=quant_config
+        )
+        self.down_proj = RowParallelLinear(
+            intermediate_size, hidden_size, bias=False, quant_config=quant_config
+        )
+        if hidden_act != "silu":
+            raise ValueError(
+                f"Unsupported activation: {hidden_act}. "
+                "Only silu is supported for now."
+            )
+        self.act_fn = SiluAndMul()
+
+    def forward(self, x):
+        gate_up, _ = self.gate_up_proj(x)
+        x = self.act_fn(gate_up)
+        x, _ = self.down_proj(x)
+        return x
+
+
+class BaiChuanAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(
+        self,
+        hidden_size: int,
+        num_heads: int,
+        position_embedding: str,
+        rope_theta: float = 10000,
+        max_position_embeddings: int = 8192,
+        quant_config: Optional[QuantizationConfig] = None,
+        layer_id: int = 0,
+    ):
+        super().__init__()
+        self.hidden_size = hidden_size
+        tensor_model_parallel_world_size = get_tensor_model_parallel_world_size()
+        tp_size = get_tensor_model_parallel_world_size()
+        self.total_num_heads = num_heads
+        assert self.total_num_heads % tensor_model_parallel_world_size == 0
+        self.num_heads = self.total_num_heads // tensor_model_parallel_world_size
+        self.head_dim = hidden_size // self.total_num_heads
+        self.postion_embedding = position_embedding
+        self.rope_theta = rope_theta
+        self.max_position_embeddings = max_position_embeddings
+        self.total_num_kv_heads = self.num_heads
+        if self.total_num_kv_heads >= tp_size:
+            # Number of KV heads is greater than TP size, so we partition
+            # the KV heads across multiple tensor parallel GPUs.
+            assert self.total_num_kv_heads % tp_size == 0
+        else:
+            # Number of KV heads is less than TP size, so we replicate
+            # the KV heads across multiple tensor parallel GPUs.
+            assert tp_size % self.total_num_kv_heads == 0
+        self.num_kv_heads = max(1, self.total_num_kv_heads // tp_size)
+
+        # pylint: disable=invalid-name
+        self.W_pack = QKVParallelLinear(
+            hidden_size,
+            self.head_dim,
+            self.total_num_heads,
+            self.total_num_heads,
+            bias=False,
+            quant_config=quant_config,
+        )
+        self.o_proj = RowParallelLinear(
+            self.total_num_heads * self.head_dim,
+            hidden_size,
+            bias=False,
+            quant_config=quant_config,
+        )
+        # Create the alibi slopes and slice them.
+        if self.postion_embedding == "ALIBI":
+            tp_rank = get_tensor_model_parallel_rank()
+            head_start = tp_rank * self.num_heads
+            head_end = (tp_rank + 1) * self.num_heads
+            alibi_slopes = _get_alibi_slopes(self.total_num_heads)
+            alibi_slopes = alibi_slopes[head_start:head_end].tolist()
+
+            scaling = self.head_dim**-0.5
+            self.attn = RadixAttention(
+                self.num_heads,
+                self.head_dim,
+                scaling,
+                num_kv_heads=self.num_kv_heads,
+                layer_id=layer_id,
+            )
+        else:
+            self.rotary_emb = get_rope(
+                self.head_dim,
+                rotary_dim=self.head_dim,
+                max_position=self.max_position_embeddings,
+                base=self.rope_theta,
+            )
+            self.scaling = self.head_dim**-0.5
+            self.attn = RadixAttention(
+                self.num_heads,
+                self.head_dim,
+                self.scaling,
+                num_kv_heads=self.num_kv_heads,
+                layer_id=layer_id,
+            )
+
+    def forward(
+        self,
+        positions: torch.Tensor,
+        hidden_states: torch.Tensor,
+        input_metadata: InputMetadata,
+    ) -> torch.Tensor:
+        qkv, _ = self.W_pack(hidden_states)
+        q, k, v = qkv.chunk(chunks=3, dim=-1)
+        if self.postion_embedding != "ALIBI":
+            q, k = self.rotary_emb(positions, q, k)
+        attn_output = self.attn(q, k, v, input_metadata)
+        output, _ = self.o_proj(attn_output)
+        return output
+
+
+class BaiChuanDecoderLayer(nn.Module):
+
+    def __init__(
+        self,
+        config: PretrainedConfig,
+        position_embedding: str,
+        layer_id: int = 0,
+        quant_config: Optional[QuantizationConfig] = None,
+    ):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        rope_theta = getattr(config, "rope_theta", 10000)
+        max_position_embeddings = getattr(config, "max_position_embeddings", 8192)
+        self.self_attn = BaiChuanAttention(
+            hidden_size=self.hidden_size,
+            num_heads=config.num_attention_heads,
+            position_embedding=position_embedding,
+            rope_theta=rope_theta,
+            layer_id=layer_id,
+            max_position_embeddings=max_position_embeddings,
+            quant_config=quant_config,
+        )
+        self.mlp = BaiChuanMLP(
+            hidden_size=self.hidden_size,
+            intermediate_size=config.intermediate_size,
+            hidden_act=config.hidden_act,
+            quant_config=quant_config,
+        )
+        self.input_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = RMSNorm(
+            config.hidden_size, eps=config.rms_norm_eps
+        )
+
+    def forward(
+        self,
+        positions: torch.Tensor,
+        hidden_states: torch.Tensor,
+        input_metadata: InputMetadata,
+        residual: Optional[torch.Tensor],
+    ) -> Tuple[torch.Tensor, torch.Tensor]:
+        # Self Attention
+        if residual is None:
+            residual = hidden_states
+            hidden_states = self.input_layernorm(hidden_states)
+        else:
+            hidden_states, residual = self.input_layernorm(hidden_states, residual)
+        hidden_states = self.self_attn(
+            positions=positions,
+            hidden_states=hidden_states,
+            input_metadata=input_metadata,
+        )
+
+        # Fully Connected
+        hidden_states, residual = self.post_attention_layernorm(hidden_states, residual)
+        hidden_states = self.mlp(hidden_states)
+        return hidden_states, residual
+
+
+class BaiChuanModel(nn.Module):
+
+    def __init__(
+        self,
+        config: PretrainedConfig,
+        position_embedding: str,
+        quant_config: Optional[QuantizationConfig] = None,
+    ):
+        super().__init__()
+        self.config = config
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = VocabParallelEmbedding(
+            config.vocab_size,
+            config.hidden_size,
+        )
+        self.layers = nn.ModuleList(
+            [
+                BaiChuanDecoderLayer(
+                    config,
+                    layer_id=i,
+                    position_embedding=position_embedding,
+                    quant_config=quant_config,
+                )
+                for i in range(config.num_hidden_layers)
+            ]
+        )
+        self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        input_metadata: InputMetadata,
+    ) -> torch.Tensor:
+        hidden_states = self.embed_tokens(input_ids)
+        residual = None
+        for i in range(len(self.layers)):
+            layer = self.layers[i]
+            hidden_states, residual = layer(
+                positions,
+                hidden_states,
+                input_metadata,
+                residual,
+            )
+        hidden_states, _ = self.norm(hidden_states, residual)
+        return hidden_states
+
+
+class BaiChuanBaseForCausalLM(nn.Module):
+    packed_modules_mapping = {
+        "W_pack": ["W_pack"],
+        "gate_up_proj": [
+            "gate_proj",
+            "up_proj",
+        ],
+    }
+    # LoRA specific attributes
+    supported_lora_modules = [
+        "W_pack",
+        "o_proj",
+        "gate_up_proj",
+        "down_proj",
+    ]
+    embedding_modules = {}
+    embedding_padding_modules = []
+
+    def __init__(
+        self,
+        config: PretrainedConfig,
+        position_embedding: str,
+        cache_config: Optional[CacheConfig] = None,
+        quant_config: Optional[QuantizationConfig] = None,
+    ):
+        super().__init__()
+
+        self.config = config
+
+        self.quant_config = quant_config
+        self.model = BaiChuanModel(config, position_embedding, quant_config)
+        self.lm_head = ParallelLMHead(
+            config.vocab_size, config.hidden_size, quant_config=quant_config
+        )
+        if self.config.tie_word_embeddings:
+            self.lm_head.weight = self.model.embed_tokens.weight
+        self.logits_processor = LogitsProcessor(config)
+        self.sampler = Sampler()
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        input_metadata: InputMetadata,
+    ) -> torch.Tensor:
+        hidden_states = self.model(input_ids, positions, input_metadata)
+        logits_output = self.logits_processor(
+            input_ids, hidden_states, self.lm_head.weight, input_metadata
+        )
+        sample_output = self.sampler(logits_output, input_metadata.sampling_info)
+
+        return sample_output, logits_output
+
+    def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
+        stacked_params_mapping = [
+            # (param_name, shard_name, shard_id)
+            ("gate_up_proj", "gate_proj", 0),
+            ("gate_up_proj", "up_proj", 1),
+        ]
+        params_dict = dict(self.named_parameters())
+        for name, loaded_weight in weights:
+            if "rotary_emb.inv_freq" in name:
+                continue
+            if name == "lm_head.weight":
+                # Unlike Baichuan, Baichuan2 normalizes the head weights.
+                # Refer to:
+                # https://huggingface.co/baichuan-inc/Baichuan2-7B-Chat/blob/84603cde5ebffb6084e476cfaeceaf0b8b91fe54/modeling_baichuan.py#L508
+                # Distinguish between Baichuan and Baichuan2 by checking the
+                # vocab size. This is suggested by
+                # https://github.com/vllm-project/vllm/pull/1022#discussion_r1325652704
+                is_baichuan2 = self.config.vocab_size == 125696
+                if is_baichuan2:
+                    loaded_weight = torch.nn.functional.normalize(loaded_weight)
+
+            for param_name, weight_name, shard_id in stacked_params_mapping:
+                if weight_name not in name:
+                    continue
+                name = name.replace(weight_name, param_name)
+                # Skip loading extra bias for GPTQ models.
+                if name.endswith(".bias") and name not in params_dict:
+                    continue
+                param = params_dict[name]
+                weight_loader = param.weight_loader
+                weight_loader(param, loaded_weight, shard_id)
+                break
+            else:
+                # Skip loading extra bias for GPTQ models.
+                if name.endswith(".bias") and name not in params_dict:
+                    continue
+                param = params_dict[name]
+                weight_loader = getattr(param, "weight_loader", default_weight_loader)
+                weight_loader(param, loaded_weight)
+
+
+class BaichuanForCausalLM(BaiChuanBaseForCausalLM):
+    """Baichuan 13B and Baichuan2 7B/13B."""
+
+    def __init__(
+        self,
+        config,
+        cache_config: Optional[CacheConfig] = None,
+        quant_config: Optional[QuantizationConfig] = None,
+    ):
+        if config.hidden_size == 4096:  # baichuan2 7b
+            super().__init__(config, "ROPE", cache_config, quant_config)
+        else:  # baichuan 13b, baichuan2 13b
+            super().__init__(config, "ALIBI", cache_config, quant_config)
+
+
+EntryClass = [BaichuanForCausalLM]