Sponsored by
APIMart
amazon / gpt-oss-120b-p-eagle
Model's Last Updated: May 02 2026
Introduction of gpt-oss-120b-p-eagle
Model Details of gpt-oss-120b-p-eagle
Model Overview
P-EAGLE is a parallel-drafting speculative decoding model that generates K draft tokens in a single forward pass. It transforms EAGLE—the state-of-the-art speculative decoding method—from autoregressive to parallel draft generation.
Model Details
The model architecture is illustrated in the following figure. Specifically, we trained a 4-layer P-EAGLE for GPT-OSS 120B as the target model.
Model Description
- Developed by: AWS
- Model type: EAGLE
- Language(s) (NLP): English
- License: Apache License 2.0
- Target model: GPT-OSS 120B
Model Sources
Training Data
Similar to nvidia/gpt-oss-120b-Eagle3-long-context : only prompts from the datasets were used for data synthesis (the original responses from GPT were not used for data synthesis) which is then used to train the P-Eagle.
Usage
To serve the checkpoint in vLLM
CUDA_VISIBLE_DEVICES=0 VLLM_USE_FLASHINFER_MOE_MXFP4_MXFP8=1 \
vllm serve openai/gpt-oss-120b --speculative-config '{"method": "eagle3", "model": "amazon/gpt-oss-120b-p-eagle", "num_speculative_tokens": 5, "parallel_drafting": true}' -tp 1 --attention-config.backend flashinfer --attention-config.force_use_trtllm true --max-num-batched-tokens 32768 --kv-cache-dtype fp8 --async-scheduling --stream-interval 20 --max-cudagraph-capture-size 4096 --no-enable-prefix-caching --port 8040 --gpu-memory-utilization 0.9 --max-num-seqs 128 --max-model-len 32768
Ciatation
@article{hui2026p,
title={P-EAGLE: Parallel-Drafting EAGLE with Scalable Training},
author={Hui, Mude and Huang, Xin and Salas, Jaime Campos and Sun, Yue and Pemberton, Nathan and Song, Xiang and Khetan, Ashish and Karypis, George},
journal={arXiv preprint arXiv:2602.01469},
year={2026}
}
Runs of amazon gpt-oss-120b-p-eagle on huggingface.co
182
Total runs
3
24-hour runs
1
3-day runs
9
7-day runs
-72
30-day runs
More Information About gpt-oss-120b-p-eagle huggingface.co Model
Url of gpt-oss-120b-p-eagle
gpt-oss-120b-p-eagle huggingface.co Url
Provider of gpt-oss-120b-p-eagle huggingface.co
Other API from amazon
Total runs:
7.0M
Run Growth:
-294.9K
Growth Rate:
-4.23%
Total runs:
1.1M
Run Growth:
308.2K
Growth Rate:
27.68%
Total runs:
1.1M
Run Growth:
-70.4K
Growth Rate:
-6.40%
Total runs:
924.6K
Run Growth:
-142.6K
Growth Rate:
-15.43%
Total runs:
716.4K
Run Growth:
120.1K
Growth Rate:
16.76%
Total runs:
422.6K
Run Growth:
-377.9K
Growth Rate:
-89.44%
Total runs:
190.5K
Run Growth:
28.9K
Growth Rate:
15.17%
Total runs:
126.1K
Run Growth:
8.0K
Growth Rate:
6.38%
Total runs:
70.1K
Run Growth:
18.5K
Growth Rate:
26.37%
Total runs:
3.8K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
529
Run Growth:
113
Growth Rate:
21.36%
Total runs:
523
Run Growth:
409
Growth Rate:
78.35%
Total runs:
429
Run Growth:
-136
Growth Rate:
-35.42%
Total runs:
409
Run Growth:
240
Growth Rate:
60.61%
Total runs:
401
Run Growth:
-5.2K
Growth Rate:
-1334.10%
Total runs:
363
Run Growth:
284
Growth Rate:
78.89%
Total runs:
317
Run Growth:
283
Growth Rate:
90.42%
Total runs:
308
Run Growth:
273
Growth Rate:
89.80%
Total runs:
291
Run Growth:
229
Growth Rate:
79.51%
Total runs:
273
Run Growth:
-740
Growth Rate:
-275.09%
Total runs:
178
Run Growth:
36
Growth Rate:
20.34%
Total runs:
126
Run Growth:
104
Growth Rate:
82.54%
Total runs:
115
Run Growth:
83
Growth Rate:
76.15%
Total runs:
92
Run Growth:
-7
Growth Rate:
-7.69%
Total runs:
82
Run Growth:
44
Growth Rate:
57.14%
Total runs:
81
Run Growth:
-214
Growth Rate:
-281.58%
Total runs:
22
Run Growth:
0
Growth Rate:
0.00%
Total runs:
16
Run Growth:
-1
Growth Rate:
-6.25%
Total runs:
15
Run Growth:
-4
Growth Rate:
-26.67%
Total runs:
13
Run Growth:
-5
Growth Rate:
-38.46%
Total runs:
12
Run Growth:
-5
Growth Rate:
-41.67%