NikitosKey/Merge-Math-Coder-7B-v2

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Dec 25, 2025Architecture:Transformer Featherless Exclusive Cold

NikitosKey/Merge-Math-Coder-7B-v2 is a 7.6 billion parameter language model created by NikitosKey, merging Qwen2.5-Math-7B-Instruct and Qwen2.5-Coder-7B-Instruct using the SLERP method. This model is designed to combine strong mathematical reasoning and coding capabilities, making it suitable for tasks requiring both logical problem-solving and code generation. It features a context length of 32768 tokens, providing ample capacity for complex prompts in its specialized domains.

Loading preview...

NikitosKey/Merge-Math-Coder-7B-v2 Overview

This model, developed by NikitosKey, is a 7.6 billion parameter language model that combines the strengths of two specialized Qwen 2.5 models: Qwen/Qwen2.5-Math-7B-Instruct and Qwen/Qwen2.5-Coder-7B-Instruct. It was created using the SLERP (Spherical Linear Interpolation) merge method, which allows for a balanced integration of the distinct capabilities of its base models.

Key Capabilities

  • Enhanced Mathematical Reasoning: Inherits robust mathematical problem-solving skills from Qwen2.5-Math-7B-Instruct.
  • Proficient Code Generation: Benefits from the coding expertise of Qwen2.5-Coder-7B-Instruct, enabling effective code generation and understanding.
  • Combined Domain Expertise: Designed to excel in tasks that require both logical mathematical thinking and practical coding solutions.
  • Extended Context Window: Supports a context length of 32768 tokens, facilitating the processing of longer and more complex prompts in both math and coding.

Good For

  • Mathematical Problem Solving: Ideal for applications requiring advanced arithmetic, algebra, calculus, and other mathematical reasoning.
  • Code Development and Assistance: Suitable for generating code snippets, debugging, explaining code, and assisting with programming tasks across various languages.
  • Hybrid Math-Coding Challenges: Particularly effective for scenarios where mathematical logic needs to be translated into code, or where code involves complex mathematical operations.
  • Educational Tools: Can be used in educational contexts for teaching programming and mathematics, or for generating practice problems and solutions.

Finetunes
1 models