backend.ai

Backend.AI now officially supports the Intel Arc Pro B70 workstation GPU, expanding its hardware coverage beyond Intel Gaudi 2 and 3 AI accelerators to include the Arc graphics lineup. This enables unified management of Intel AI hardware across environments, from data center Gaudi to workstation-class Arc Pro, within a single platform. Backend.AI supports diverse GPUs and accelerators with an intuitive interface and session-based resource management, simplifying AI model development, training, and deployment. With 32GB memory and high throughput, Arc Pro B70 is optimized for agentic AI workloads, improving concurrency and KV cache efficiency.

Lablup wrapped up AI EXPO KOREA 2026 at booth F04. A three-day recap, from Backend.AI's AI infrastructure orchestration to AI:GO running models and autonomous agents on a laptop.

Lablup has joined the Python Software Foundation (PSF) as a Participating Sponsor, continuing its support for the open source ecosystem.

Lablup open-sources mlxcel, an AI inference engine optimized for Apple Silicon (M1 to M5) and NVIDIA CUDA. Built in pure Rust with no Python runtime, it delivers 119% of mlx-lm's decode throughput and supports 80+ model architectures.

A rundown of the major changes in Backend.AI FastTrack 3 25.18.

Backend.AI 25.15 LTS is here, with broad system and UX optimization that strengthens reliability and scalability for large-scale AI training and deployment.

We provide a step-by-step tutorial on how to install the open-source version of Backend.AI on NVIDIA DGX Spark and easily launch a model.

The AI accelerator market is diversifying across NVIDIA, AMD, Intel, and domestic NPUs, enabling heterogeneous GPU operations. Backend.AI solves software stack and performance challenges by separating resources into groups and applying workload-specific policies to maximize each accelerator's strengths.

Vision-language models detect real emergencies reasonably well, but they often overreact by misclassifying safe scenes as dangerous. This post shares results from the VERI benchmark, which measures this limitation using visually similar emergency and safe scenes with different meanings.

The Upstage consortium, with Lablup as infrastructure partner, passed Phase 1 of Korea's Sovereign AI Foundation Model program. Team members share the behind-the-scenes story.