MMOPD/Qwen3-4B-OT3-2ep

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 14, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

MMOPD/Qwen3-4B-OT3-2ep is a 4 billion parameter Qwen3-Base model fine-tuned by MMOPD on the OpenThoughts3-1.2M dataset for two epochs. This model is specifically designed as a "thinking" model, generating a detailed thought process before each answer. It excels in complex reasoning tasks across math, code, and science, serving as a general reasoning student for further domain-specific distillation. The model supports a context length of 32768 tokens and is optimized for tasks requiring explicit step-by-step reasoning.

Loading preview...

Overview

MMOPD/Qwen3-4B-OT3-2ep is a 4 billion parameter model based on Qwen3-4B-Base, developed by MMOPD. It has been fine-tuned for two epochs on the OpenThoughts3-1.2M dataset, which comprises 1.2 million examples of long-chain-of-thought reasoning traces covering mathematics, coding, and scientific problems. This model is characterized as a "thinking" model, meaning every generated answer begins with a <think> block, explicitly outlining its reasoning process.

Key Capabilities

  • Enhanced Reasoning: Demonstrates strong performance in complex reasoning tasks across math, code, and science, as evidenced by its benchmarks on AIME, LiveCodeBench, IFEval, and IFBench.
  • Explicit Thought Process: Generates a detailed thought block (<think> ... </think>) before providing an answer, making its reasoning transparent.
  • Foundation for Specialization: Serves as a general reasoning student, intended as a starting point for further domain-specific fine-tuning within the MMOPD project (e.g., medical, law, finance).
  • Long Context Support: Trained with a sequence length of 16,384 tokens and supports a maximum generation budget of 32,768 tokens.

Performance Highlights

Evaluations show improved performance over its 1-epoch counterpart (MMOPD/Qwen3-4B-OT3-1ep). For instance, it achieves 66.3% on AIME24, 51.7% on LiveCodeBench v6, and 51.0% on IFEval. Domain-specific benchmarks include 69.8% on MedQA and 58.3% on FinQA.

When to Use This Model

This model is particularly suitable for use cases requiring:

  • Complex Problem Solving: Ideal for applications needing robust reasoning in technical domains.
  • Explainable AI: When understanding the model's step-by-step thought process is crucial.
  • Foundation for Customization: Developers looking for a strong reasoning base to further fine-tune for specific domain expertise.

Finetunes
1 models