GammaAGI/AIS-Gamma-Nemotron-Reasoning-Code-TIES-32B

TEXT GENERATIONConcurrency Cost:2Model Size:32.8BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:May 9, 2026Architecture:Transformer0.0K Cold

The AIS-Gamma-Nemotron-Reasoning-Code-TIES-32B is a 32.8 billion parameter language model developed by GAMMAai, utilizing a TIES merge of NVIDIA's OpenReasoning-Nemotron-32B and OpenCodeReasoning-Nemotron-32B. Built on the Qwen2ForCausalLM architecture, this model is specifically optimized for advanced reasoning and code generation tasks. It offers a 32768-token context length, making it suitable for complex problem-solving and extensive code analysis.

Loading preview...

AIS Nemotron Reasoning+Code TIES 32B Overview

This model, part of the GAMMAai research and model-fusion program, is a 32.8 billion parameter language model designed for enhanced reasoning and code generation capabilities. It leverages a novel TIES merge method to combine the strengths of two specialized NVIDIA Nemotron base models.

Key Characteristics & Merge Details

  • Architecture: Based on the robust Qwen2ForCausalLM architecture.
  • Merge Method: Utilizes a TIES merge via mergekit to integrate functionalities.
  • Base Models: Merges nvidia/OpenReasoning-Nemotron-32B and nvidia/OpenCodeReasoning-Nemotron-32B, indicating a dual focus on logical deduction and programming tasks.
  • Size: The merged model is approximately 32.8 billion parameters, resulting in a 62 GB BF16 footprint.
  • Context Length: Supports a substantial context window of 32768 tokens, beneficial for handling intricate problems and large codebases.

Primary Use Cases

  • Complex Reasoning: Excels in tasks requiring logical inference and problem-solving.
  • Code Generation & Analysis: Optimized for generating high-quality code and assisting with code-related reasoning.

This model represents an effort to create a universal AI by merging over 70 models, with this specific iteration focusing on combining advanced reasoning with strong coding abilities.