diff --git a/README.md b/README.md
index 3875893a..48f0d510 100644
--- a/README.md
+++ b/README.md
@@ -180,7 +180,7 @@ A more accessible, comprehensive, and efficient toolkit for large model compress
@@ -341,13 +341,19 @@ For more detaileds, please refer to the [Deployment Documentation](https://angel
### 1. Speculative Decoding
-#### 1.1 Qwen3 Series Models
+We evaluated the Eagle3 model trained by AngelSlim on tasks including code generation, mathematical reasoning, instruction following, text generation, and multimodal understanding using vLLM. The inference acceleration and context length performance of our trained model under the settings of num_speculative_tokens = 2 or 4 are presented as follows.
+
+
+
+
+
+
+
+
-**vLLM v0.11.2 Benchmark Results**
+#### 1.1 Qwen3 Series Models
-We report benchmark results of the Qwen3 series models using the Eagle3 speculative decoding algorithm across multiple evaluation suites, including **MT-bench**, **HumanEval**, **GSM8K**, and **Alpaca**.
-All experiments were conducted on a single NVIDIA H20 GPU with the configuration:
-**tp=1, ep=1, num_speculative_tokens=2, batch_size=1, output_len=1024**.
+Benchmark results for Qwen3 series models using Eagle3 speculative decoding on vLLM (v0.11.2) across **MT-bench**, **HumanEval**, **GSM8K** and **Alpaca**, using a single NVIDIA H20 GPU (**tp=1, ep=1, num_speculative_tokens=2, batch_size=1, output_len=1024**).
@@ -379,7 +385,7 @@ All experiments were conducted on a single NVIDIA H20 GPU with the configuration
378.86
1
378.38
1
390.53
1
-
318.05
1
+
381.05
1
Eagle3
@@ -387,7 +393,7 @@ All experiments were conducted on a single NVIDIA H20 GPU with the configuration
653.29
2.19
680.1
2.2
621.44
2.17
-
642.93
2.18
+
642.93
2.17
@@ -483,6 +489,251 @@ All experiments were conducted on a single NVIDIA H20 GPU with the configuration
+#### 1.2 VLM Models
+
+##### 1.2.1 Qwen3-VL Series Models
+
+Benchmark results for Qwen3-VL series models using Eagle3 speculative decoding on vLLM (v0.12.0) across language and multimodal tasks, using a single NVIDIA H20 GPU (**tp=1, ep=1, num_speculative_tokens=4, batch_size=1, output_len=1024**).
+
+
+
+
Model
+
Method
+
GSM8K
+
Alpaca
+
HumanEval
+
MT-bench
+
MATH-500
+
MMMU
+
MMStar
+
+
+
+
+
+
throughput (tokens/s)
+
accept length
+
throughput (tokens/s)
+
accept length
+
throughput (tokens/s)
+
accept length
+
throughput (tokens/s)
+
accept length
+
throughput (tokens/s)
+
accept length
+
throughput (tokens/s)
+
accept length
+
throughput (tokens/s)
+
accept length
+
+
+
Qwen3-VL-2B-Instruct
+
Vanilla
+
348.55
+
1
+
350.9
+
1
+
346.07
+
1
+
346.31
+
1
+
82.96
+
1
+
83.27
+
1
+
81.63
+
1
+
+
+
Eagle3
+
511.52
+
2.11
+
560.55
+
2.26
+
826.01
+
3.39
+
555.22
+
2.29
+
163.09
+
2.57
+
154.18
+
2.55
+
139.73
+
2.31
+
+
+
Qwen3-VL-4B-Instruct
+
Vanilla
+
212.87
+
1
+
213.24
+
1
+
211.69
+
1
+
212.1
+
1
+
67.96
+
1
+
65.88
+
1
+
67.75
+
1
+
+
+
Eagle3
+
415.29
+
2.57
+
372.89
+
2.26
+
459.37
+
2.82
+
382.33
+
2.34
+
141.87
+
2.72
+
104.44
+
2.05
+
107.07
+
2.1
+
+
+
Qwen3-VL-30B-A3B-Instruct
+
Vanilla
+
179.94
+
1
+
184.6
+
1
+
168.68
+
1
+
180.57
+
1
+
31.08
+
1
+
31.51
+
1
+
30.93
+
1
+
+
+
Eagle3
+
281.93
+
2.82
+
241.42
+
2.13
+
223.05
+
2.57
+
240.47
+
2.19
+
75.31
+
2.79
+
48.47
+
1.78
+
52.57
+
1.94
+
+
+
+##### 1.2.2 HunyuanOCR Model
+
+Benchmark results for HunyuanOCR using Eagle3 speculative decoding on vLLM (v0.13.0) across OCR tasks, using a single NVIDIA H20 GPU (**tp=1, ep=1, num_speculative_tokens=4, batch_size=1, output_len=1024**).
+
+
+
+
Model
+
Method
+
OCR-Bench-Internal
+
+
+
+
+
+
+
throughput (tokens/s)
+
accept length
+
+
+
Hunyuan-OCR
+
Vanilla
+
71.21
+
1
+
+
+
+
Eagle3
+
120.75
+
2.2
+
+
+
+
+#### 1.3 Audio Models
+
+##### 1.3.1 Qwen2-Audio Model
+
+Benchmark results for Qwen2-Audio using Eagle3 speculative decoding on vLLM (v0.12.0) across **[LibriSpeech](https://www.openslr.org/12)** dataset, using a single NVIDIA H20 GPU (**tp=1, ep=1, num_speculative_tokens=4, batch_size=1, output_len=1024**).
+
+
+
+
Model
+
Method
+
LibriSpeech
+
+
+
+
+
+
throughput (tokens/s)
+
accept length
+
+
+
Qwen2-Audio-7B-Instruct
+
Vanilla
+
78.76
+
1
+
+
+
+
Eagle3
+
146.66
+
3.51
+
+
+
+
+##### 1.3.2 Fun-CosyVoice3 Model
+
+Benchmark results for Fun-CosyVoice3 using Eagle3 speculative decoding across **[LibriTTS](https://www.openslr.org/60/)** dataset, using a single NVIDIA H20 GPU (**tp=1, ep=1, num_speculative_tokens=4, batch_size=1, output_len=1024**).
+
+
+
+
Model
+
Method
+
LibriTTS
+
+
+
+
+
+
throughput (tokens/s)
+
accept length
+
+
+
Fun-CosyVoice3
+
Vanilla
+
-
+
1
+
+
+
+
Eagle3
+
-
+
1.96
+
+
+
+
+> Adapted for Transformers backend inference, only displays accept length.
+
### 2. Quantization
The performance test results for selected models are shown below. For the complete benchmark, refer to the [Benchmark documentation](https://angelslim.readthedocs.io/zh-cn/latest/performance/quantization/benchmarks.html)
diff --git a/README_cn.md b/README_cn.md
index 8b9d262e..e5ab1c07 100644
--- a/README_cn.md
+++ b/README_cn.md
@@ -181,7 +181,7 @@
+