sayeed105236/CuttyMOA-1.3

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 5, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

CuttyMOA-1.3 is a 27 billion parameter Mixture-of-Experts (MoE) AI assistant developed by MD ABU SAYEED, based on the Qwen3.8-27B architecture. It features a 262,144 token context window and is fine-tuned for complex reasoning, code generation, cybersecurity intelligence, and agentic tool-calling. This model is designed as the core reasoning engine for the CUTEADMOA platform, excelling in multi-step tasks and long-horizon problem-solving.

Loading preview...

CuttyMOA-1.3: CUTEADMOA's Flagship AI Assistant

CuttyMOA-1.3, developed by MD ABU SAYEED and powered by DOTPROGRAMMERS, is a 27 billion parameter Mixture-of-Experts (MoE) model built on the Qwen3.8-27B architecture. It boasts an extensive 262,144 token context window and an 8,192 token maximum output, making it suitable for processing large documents and codebases. The model is fine-tuned using LoRA with a focus on diverse enterprise applications.

Key Capabilities

  • Advanced Reasoning & Chat: Excels in long-horizon multi-step tasks, instruction following, and summarization.
  • Code Generation & Analysis: Strong performance in code generation, review, debugging, and SQL, with code being the largest corpus category.
  • Security Intelligence: Trained on 17 cybersecurity intents for severity classification.
  • Multilingual Research: Supports 24 intent categories across multiple languages (EN, BN, HI, AR, ZH, JA, KO, FR, DE, ES).
  • Agentic Tool-Calling: Trained on various function-calling corpora (BFCL, xLAM, Glaive, Hermes).
  • Data Extraction: Capable of extraction planning and JSON/CSV schema design for scraping-adjacent tasks.
  • Vision-Language: Inherits native image/video understanding from its Qwen3.8 base, though text performance was the primary training focus.

Training & Integration

The model was fine-tuned on a proprietary CUTEADMOA corpus combining 204-verified domain datasets and live-harvested agent pairs, with a unique identity injection for self-identification. It is integrated into the CUTEADMOA platform, offering a live playground and an OpenAI-compatible API. A smaller 4B sibling, CuttyMOA-1.3-Small, is also available for local deployment.

Finetunes
1 models