EdgeRunner AI, the leader in military-specific, on-device artificial intelligence, today announced EdgeRunner 20B, a new research model that matches GPT-5 level performance on four military test sets ...

Defense tech startup becomes first in the world to operationalize GPT-5 class AI capabilities entirely on-device for the military in DDIL environments
SEATTLE: EdgeRunner AI, the leader in military-specific, on-device artificial intelligence, today announced EdgeRunner 20B, a new research model that matches GPT-5 level performance on four military test sets while running completely air-gapped on-device in Denied, Disrupted, Intermittent, and Limited (DDIL) environments. The company released a research paper detailing the full results.
“Two years ago many believed GPT-4 level performance on a commodity device was impossible due to compute, memory, and power constraints inherent to edge hardware,” said Tyler Saltsman, Co-Founder & CEO at EdgeRunner AI. “We overcame those limits-achieving GPT-5 level performance entirely on-device-and we are the first in the world to operationalize this for military use cases in contested and denied environments."
EdgeRunner 20B will be integrated into EdgeRunner AI’s state-of-the-art Generative AI platform, which delivers occupation and mission-specific AI agents air-gapped on-device. Specifically developed for the warfighter, this technology offers both defense and enterprise users a hyper-personalized assistant powered by AI agents that functions independent of internet connectivity for increased privacy and security.
The platform is powered by state-of-the-art Small Language Models (SLMs) of various parameter sizes, including EdgeRunner-Light, EdgeRunner-Medium, and EdgeRunner-Heavy, purpose-built for military operations. Optimized for secure, on-device deployment across AI PCs, mobile devices, and edge computing environments, the platform delivers high-performance AI capabilities without requiring internet access. EdgeRunner supports a wide range of functions, including chat, Q&A, summarization, language translation, transcription, code generation, multi-document RAG (PDF, Word, PowerPoint), text-to-speech (TTS), and speech-to-text (STT). Military Occupational Specialty (MOS)-specific adapters tailored for roles such as logistics, combat medic, and combat arms ensure responses are contextually accurate and operationally relevant for each user.
About EdgeRunner AI
EdgeRunner AI is the leader in military-specific, on-device artificial intelligence, offering the world’s fastest on-device military AI platform, engineered for Denied, Disrupted, Intermittent, and Limited (DDIL) environments. With cutting-edge LLMs, intelligent agents, and computer vision systems, EdgeRunner AI delivers warfighters AI-driven speed, precision, and autonomy-without the cloud. For more information, visit: https://www.edgerunnerai.com
Fonte: Business Wire
Alaa Abdul Nabi, Vice President, Sales International at RSA presents the innovations the vendor brings to Cybertech as part of a passwordless vision for…
G11 Media's SecurityOpenLab magazine rewards excellence in cybersecurity: the best vendors based on user votes
Always keeping an European perspective, Austria has developed a thriving AI ecosystem that now can attract talents and companies from other countries
Successfully completing a Proof of Concept implementation in Athens, the two Italian companies prove that QKD can be easily implemented also in pre-existing…
Reply [EXM, STAR: REY] announced that it has achieved the Amazon Web Services (AWS) Agentic AI Specialization, a new category within the AWS AI Competency.…
Tecnotree, a global digital platform and services leader for AI, 5G, and cloud-native technologies, has won the CX Catalyst award for Impact at The Fast…
#AGI--Integral AI, a global leader in the development of embodied AGI, today announced the successful testing of the world’s first AGI-capable model.…
The "India Existing & Upcoming Data Center Portfolio" database has been added to ResearchAndMarkets.com's offering. India hosts around 132 existing…