Back to News
DeepSeek•June 27, 2026
DeepSpec
Open Source
open-source
speculative decoding
DeepSeek
AI research
### TL;DR
DeepSpec is an open-source, full-stack codebase developed by DeepSeek for training and evaluating speculative decoding algorithms. It encompasses data preparation utilities, draft model implementations, training code, and evaluation scripts, facilitating the development and assessment of draft models for speculative decoding tasks.
Key Insights & Metrics
Pricing
Free
Cost structure
Version
1.0
Current release version
Hardware
Single 8-GPU node; approximately 38 TB of storage for data preparation
Compute requirements
Category
Open Source
Licensing model
Region
China
Primary region
Key Features
- Open-source codebase for speculative decoding algorithms
- Includes data preparation utilities, draft model implementations, training code, and evaluation scripts
- Supports algorithms like DSpark, DFlash, and Eagle3
- Released under the MIT License
- Requires a single 8-GPU node and approximately 38 TB of storage for data preparation
Discussion
0
Upvotes
0
Downvotes
0 reviews
Sign in to leave a review
Reviews
No reviews yet. Be the first to review!