GenueAI/Geode-Onyx-2

TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kPublished:Apr 19, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

Genue AI's Geode Onyx 2 is a 3-billion parameter conversational AI model, fine-tuned from Qwen 2.5 3B Instruct with a 32k context length. It is designed for balanced logic and personality, excelling in mathematical reasoning, general knowledge, and conversational generalization. The model also includes training for HTML/CSS/JavaScript code generation and anti-hallucination capabilities, making it suitable for diverse interactive applications.

Loading preview...

Geode Onyx 2: A Balanced Conversational AI

Genue AI's Geode Onyx 2 is a 3-billion parameter conversational model, part of the second generation of the Geode family. Fine-tuned from the Qwen 2.5 3B Instruct base model using LoRA, Onyx 2 is engineered to provide a balance of logical reasoning and engaging personality.

Key Capabilities

  • Conversational AI: Designed for natural dialogue patterns and general conversational generalization.
  • Reasoning: Strong performance in mathematical reasoning (arithmetic, algebra, word problems) and physics problems.
  • Knowledge: Possesses broad general knowledge and specific understanding of the Genue AI ecosystem.
  • Code Generation: Includes training for HTML/CSS/JavaScript code generation tasks.
  • Anti-Hallucination: Incorporates training to properly handle unknown information, reducing factual errors.
  • Efficiency: Achieves its capabilities within a 3B parameter footprint, making it suitable for local deployment.

Good For

  • Interactive Chatbots: Ideal for applications requiring balanced responses, combining factual accuracy with a conversational tone.
  • Educational Tools: Can assist with mathematical and scientific problem-solving.
  • Developer Assistants: Capable of generating basic web development code snippets.
  • General-Purpose AI: A versatile model for tasks requiring a blend of reasoning, knowledge, and conversational fluency.