prithivMLmods/Q3.5-9B-DS-v4-Flash-v2.0

VISIONConcurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 1, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

prithivMLmods/Q3.5-9B-DS-v4-Flash-v2.0 is a 9-billion parameter language model built upon Qwen/Qwen3.5-9B, developed by prithivMLmods. This model is specifically trained through a multi-stage pipeline using approximately 3K long-context DeepSeek V4 Flash reasoning traces to enhance long-form reasoning, mathematical problem-solving, scientific analysis, and instruction-following capabilities. It features a 32,768-token context length and is optimized for research and local deployment in reasoning-intensive tasks.

Loading preview...

Overview

prithivMLmods/Q3.5-9B-DS-v4-Flash-v2.0 is a 9-billion parameter language model based on the Qwen/Qwen3.5-9B foundation. It has undergone a multi-stage supervised fine-tuning (SFT) process, incorporating approximately 3,000 long-context DeepSeek V4 Flash reasoning traces, alongside other high-quality reasoning datasets. This training regimen is designed to significantly improve the model's performance in complex reasoning tasks, instruction following, and analytical problem-solving.

Key Capabilities

  • Enhanced Reasoning: Specialized training for long-form reasoning, mathematical problem-solving, and scientific analysis.
  • Long Context Window: Supports a maximum sequence length of 32,768 tokens, facilitating complex, multi-step reasoning over extensive inputs.
  • Instruction Following: Strengthened ability to understand and execute intricate instructions.
  • Efficient Deployment: A 9B parameter model suitable for local inference and research environments.

Intended Use Cases

  • Reasoning Research: Ideal for studying advanced reasoning techniques and multi-stage training methodologies.
  • Mathematical & Scientific Problem Solving: Excels in tasks requiring structured analytical thought.
  • Coding Assistance: Improves code understanding and generation through its long-context reasoning abilities.
  • Instruction Following Evaluation: Useful for assessing and refining instruction-following performance.