Accuracy preserved: Matches base model on HumanEval (96.91% pass@1)
Shorter reasoning: ~24% reduction in chain-of-thought length
Higher efficiency:+31.6% more correct solutions per token
⚠️Trade-off: −1.24% on HumanEval+ −7.2% on MMLU-Pro (Indicating reduced general knowledge reasoning performance)
⚠️Note: Due to the scope of SFT data and training focus, the model may underperform the base model on certain tasks requiring long-context understanding or more complex multi-step reasoning. The efficiency and accuracy results reported here are based solely on the HumanEval and HumanEval+ benchmarks. Thank you for your understanding.
HCaJnUQaoAAaMIc
💡 Model Introduction
Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2 is the second iteration of this reasoning-focused Qwen3.5-27B fine-tune, built to drastically improve the efficiency of chain-of-thought generation, unlocking highly substantial gains in reasoning speed and cost-reduction while actually increasing absolute accuracy.
Compared with the earlier version, v2 was trained with 14,000 Claude 4.6 Opus-style general reasoning samples, with a stronger emphasis on transferring concise, reusable reasoning patterns rather than only maximizing raw benchmark scores. The goal of v2 is not simply to make the model "think more," but to help it think more economically: reducing unnecessarily long internal chains, avoiding verbose over-analysis on easy problems, and massively improving the reasoning-cost-to-quality ratio while beating the baseline's benchmark correctness.
A key design choice in v2 is that the distillation data is primarily general-domain reasoning data—specifically focused on mathematics, word problems, logical deduction, and a balanced mix of general knowledge and instructions—rather than specialized code-heavy supervision. Consequently, HumanEval and HumanEval+ are employed here to evaluate cross-task generalization and capability transfer, rather than serving as direct optimization targets. High performance on these benchmarks, despite the lack of code-centric training, confirms that the model's reasoning scaffold has become more robust and transferable, proving that fundamental reasoning logic can effectively power specialized tasks like programming.
HumanEval Benchmark Analysis 🪐
The raw evaluation outputs for both models were independently cleaned, verified, and aggregated using GPT-5.4-Pro-Thinking. The final comparative results are based on these standardized and curated outputs. To ensure reliability, all results were further cross-checked and consolidated through two rounds of independent validation using Claude-4.6-Opus-Thinking.
-All evaluations were conducted in an inference environment based on Unsloth + vLLM (BF16) to ensure consistent and efficient execution conditions.
Screenshot 2026-03-20 at 3.54.26 PM
Screenshot 2026-03-20 at 3.54.51 PM
Screenshot 2026-03-20 at 3.58.30 PM
Screenshot 2026-03-20 at 3.55.49 PM
Screenshot 2026-03-20 at 3.56.01 PM
Screenshot 2026-03-20 at 3.56.18 PM
🗺️ Training Pipeline Overview
text
1Base Model (Qwen3.5-27B)
2 │
3 ▼
4Qwen3.5-27B fine-tuned with Unsloth
5 │
6 ▼
7Supervised Fine-Tuning (SFT) + LoRA
8(Response-Only Training masked on "<|im_start|>assistant\n<think>")
9 │
10 ▼
11Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2
🧠 Example of Learned Reasoning Scaffold（Example）
The model includes targeted optimizations addressing Qwen3.5’s tendency toward excessive transitional or repetitive reasoning on simple queries. Through deep distillation and structural imitation of Claude-4.6-Opus reasoning chains, the model adopts a more efficient structured thinking pattern: “Let me analyze this request carefully: 1..2..3...”.
This streamlined reasoning paradigm significantly reduces redundant cognitive loops while preserving deep analytical capacity, resulting in substantially improved inference efficiency.
text
1Let me analyze this request carefully:
231. Identify the core objective of the problem.
42. Break the task into clearly defined subcomponents.
53. Evaluate constraints and edge cases.
64. Formulate a step-by-step solution plan.
75. Execute the reasoning sequentially and verify consistency.
8 .
9 .
10 .
📚 All Datasets Used
The dataset consists of high-quality, filtered reasoning distillation data:
Additional curated reasoning samples designed to strengthen structured step-by-step problem solving and improve reasoning diversity.
⚠️ Limitations & Intended Use
Hallucination Risk: While reasoning is strong, the model remains an autoregressive LLM; external facts provided during the thinking sequence may occasionally contain hallucinations if verifying real-world events.
Intended Scenario: Best suited for offline analytical tasks, coding, math, and heavy logic-dependent prompting where the user needs to transparently follow the AI's internal logic.
This model is a test version intended solely for learning and demonstration purposes, and is for academic research and technical exploration use only.
🙏 Acknowledgements
Significant thanks to the Unsloth AI team for making rapid fine-tuning of large LLM models accessible. Additionally, we acknowledge Qwen internally, and the open-source community developers producing exceptional distilled datasets.
📖 Citation
If you use this model in your research or projects, please cite:
