OrionLLM/GRM-2.6-Plus

Hugging Face
VISIONConcurrency Cost:2Model Size:27BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Apr 23, 2026License:apache-2.0Architecture:Transformer0.1K Open Weights Warm

GRM-2.6-Plus by OrionLLM is a 27-billion parameter reasoning model built on the Qwen3.6 architecture, optimized for general-purpose AI and difficult, high-complexity tasks. It focuses on structured reasoning to deliver accurate, coherent, and reliable responses, excelling in advanced problem-solving, coding, and agentic applications. The model is designed for elite-level reasoning performance relative to its size, making it practical for advanced local and research-oriented use. It achieves strong benchmark results, including 86.8 on MMLU-Pro and 84.8 on LiveCodeBench v6, outperforming several comparable models.

Loading preview...

OrionLLM/GRM-2.6-Plus: A 27B Reasoning Model

GRM-2.6-Plus, developed by OrionLLM, is a 27-billion parameter model built on the Qwen3.6 architecture, specifically optimized for general-purpose AI and complex reasoning tasks. It emphasizes structured reasoning to provide highly accurate and consistent responses, making it suitable for demanding problems.

Key Capabilities

  • Elite-Level Reasoning: Optimized for difficult workloads, offering clarity, consistency, and strong step-by-step problem-solving.
  • High Performance for Size: Delivers excellent capability for a 27B-parameter model, balancing intelligence with practical deployment on consumer and workstation hardware.
  • Advanced Coding & Agentic Use: Well-suited for code generation, structured problem-solving, tool-style workflows, and local agentic applications.

Performance Highlights

GRM-2.6-Plus demonstrates strong performance across various benchmarks, often outperforming models in similar size classes. Notable scores include:

  • MMLU-Pro: 86.8
  • MMLU-Redux: 94.2
  • C-Eval: 92.0
  • LiveCodeBench v6: 84.8
  • SWE-bench Verified: 77.7

Good For

  • Users requiring powerful reasoning capabilities for complex problems.
  • Developers working on advanced coding tasks and agentic applications.
  • Local deployment on capable consumer or workstation hardware where efficiency and strong performance are crucial.