OrionLLM/GRM-2.6-Air-Opus
OrionLLM/GRM-2.6-Air-Opus is a 1.2 billion parameter language model developed by OrionLLM, fine-tuned from LiquidAI/LFM2.5-1.2B-Thinking. It specializes in Opus-style structured reasoning, distilled from Claude Opus traces to enhance problem-solving. This compact, edge-friendly model is designed for lightweight local and on-device deployment, offering improved reasoning capabilities for harder tasks.
Loading preview...
GRM-2.6-Air-Opus: Compact Reasoning Model
GRM-2.6-Air-Opus, developed by OrionLLM, is a 1.2 billion parameter model fine-tuned from LiquidAI/LFM2.5-1.2B-Thinking. It is specifically trained on Claude Opus-style reasoning traces to improve its structured thinking and performance on complex problems. This model is designed to be a compact, "Air" entry in the GRM-2.6 family, making it suitable for edge and on-device applications where a smaller footprint and faster operation are crucial.
Key Capabilities
- Opus-Style Reasoning: Distills the clear, organized reasoning patterns found in Claude Opus, but within a 1.2B parameter model.
- Edge-Ready Deployment: Built on LFM2.5-1.2B-Thinking, making it well-suited for low-resource environments and on-device use.
- Enhanced Problem-Solving: Targeted training on difficult reasoning traces leads to better performance on challenging tasks compared to its base model.
Performance Highlights
GRM-2.6-Air-Opus focuses on delivering practical, lightweight reasoning. Benchmarks indicate a 31.99 score on MMLU, demonstrating its knowledge and STEM capabilities within its compact size.