Open-weight, mission-specific models deployed locally offer greater performance, efficiency, and control for critical national security use cases

PITTSBURGH: EdgeRunner AI, the leader in military-specific, on-device artificial intelligence, and AI2C, the U.S. Army’s Artificial Intelligence Integration Center, today announced EdgeRunner-Camo, a new open-weight Large Language Model (LLM) from EdgeRunner AI developed with AI2C specifically for the U.S. Army.

Fine-tuned on Army-specific data, including from AI2C’s CamoGPT, the new EdgeRunner-Camo model is specifically tailored for Army users and designed for running locally in air-gapped environments without reliance upon third-party cloud-hosted proprietary models.

“We are excited to partner with AI2C to develop Army-specific LLMs for the warfighter,” said Tyler Saltsman, Co-Founder & CEO at EdgeRunner AI. “The Department of War and our national security partners require open-weight models fine-tuned on mission-specific data deployable in air-gapped environments away from the cloud. EdgeRunner is uniquely positioned to be the DoW’s key partner for building mission-specific models.”

This collaboration demonstrates how EdgeRunner AI partners with the Department of War (DoW) to train open-weight, mission-specific LLMs for national security use cases that can outperform frontier-level models while deploying locally for increased data security, privacy, and control. Furthermore, running models locally mitigates the need to worry about token usage and third-party API costs, delivering unlimited and unmetered intelligence to the warfighter at the tactical edge.

The EdgeRunner-Camo model was developed through EdgeRunner AI’s proprietary model development pipeline, utilizing Impact Level 5 (IL5) training environments when handling Controlled Unclassified Information (CUI) data. The model excels at general military tasks and Army-specific use cases, reducing error rates by up 37% on 6 new Army-specific benchmarks created as part of this partnership. Detailed results and methodology are available in a separate report from EdgeRunner AI and AI2C.

These models form the basis of an ongoing collaboration with AI2C to develop best-in-class LLMs for the Army and will be deployed into the EdgeRunner AI platform and other systems accessible to national security partners.

About EdgeRunner AI

EdgeRunner AI is redefining battlefield intelligence with the world’s best on-device AI platform purpose-built for modern warfare. Designed for Denied, Disrupted, Intermittent, and Limited (DDIL) environments, EdgeRunner AI delivers mission-critical AI directly to the tactical edge where connectivity is contested and decisions can’t wait. Powered by state-of-the-art LLMs, autonomous agents, and next-generation computer vision systems, EdgeRunner AI equips warfighters with real-time, AI-driven intelligence, enabling faster decisions, greater operational autonomy, and unmatched mission effectiveness in the world’s most demanding environments. Learn more at: https://www.edgerunnerai.com

About AI2C

Located at Carnegie Mellon University in Pittsburgh, Pennsylvania, the Army Artificial Intelligence Integration Center (AI2C) develops and integrates AI capabilities Army-wide to enhance lethality, achieve operational overmatch, and empower Soldiers and units to compete, fight, and win future conflicts. Learn more at: https://ai2c.army.mil/

Fonte: Business Wire