fableforge-ai/NEXUS-Coder

TEXT GENERATIONConcurrent Unit Cost:1Model Size:1.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 5, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

fableforge-ai/NEXUS-Coder is a 1.5 billion parameter language model based on Qwen2.5-1.5B-Instruct, developed by FableForge. It is specifically fine-tuned as a domain specialist for software engineering, code generation, debugging, and shell scripting. This lightweight model achieves a 94% score on a 30-prompt evaluation covering domain knowledge, uncensored responses, reasoning, and tool use, demonstrating strong performance in its specialized areas.

Loading preview...

NEXUS-Coder: A Specialized Code LLM

NEXUS-Coder, developed by FableForge, is a 1.5 billion parameter model built upon the Qwen2.5-1.5B-Instruct architecture. It is explicitly designed as a domain specialist for software engineering tasks, including code generation, debugging, and shell scripting.

Key Capabilities & Performance

  • Specialized Code Expertise: Optimized for software development, offering capabilities in code generation, debugging, and shell command interpretation.
  • High Benchmark Score: Achieved a total score of 141/150 (94%) in a 30-prompt evaluation. This evaluation covered:
    • Domain Knowledge: 4.8/5
    • Uncensored (Domain & General): 4.8/5
    • Reasoning: 5.0/5
    • Tool Use: 4.0/5
  • Lightweight & Efficient: Despite its small size, it demonstrates performance comparable to larger models in its specialized domain.
  • Uncensored Responses: Provides uncensored outputs relevant to its domain and general queries.
  • Part of the NEXUS Family: Belongs to a series of six specialized, lightweight models, each focusing on a distinct domain like cybersecurity, medicine, or finance.

Ideal Use Cases

  • Code Generation: Generating code snippets or full functions in various programming languages.
  • Debugging Assistance: Helping identify and resolve issues in existing code.
  • Shell Scripting: Assisting with the creation and understanding of shell commands and scripts.
  • Educational Tools: Supporting learning and development in programming and software engineering.
  • Resource-Constrained Environments: Suitable for deployment where computational resources are limited, such as on local machines via Ollama.