AI Briefing#nvidia-blackwell
The latest AI and developer news about #nvidia-blackwell, with the original source and a short summary.
The latest AI and developer news about #nvidia-blackwell, with the original source and a short summary.
A forensic report identifies NVIDIA Blackwell hardware signatures in the initial weights of six Chinese models, including DeepSeek-V4-Flash and Kimi K3, indicating access to export-controlled chips.
The Triton Low-Level Extensions (TLX) framework enables explicit warp-group orchestration, delivering up to 2x forward speedup in GDPA kernels and reducing RecSys inference latency by up to two-thirds.
New products include Ultra-fast inference on NVIDIA Blackwell, Nexus Desktop for model routing, and Fireworks Sovereign for on-premises deployment.
DeepSeek has released the offline inference code for DeepInfer, its inference framework for V4.1-Flash.
A checkpoint quantizing Qwen3.8-27B to NVFP4 using the QUASAR QAT algorithm has been released, maintaining BF16-level performance.
Firebird has launched the largest NVIDIA-based AI factory in the CIS region in Armenia.
It implements a block-diagonal attention kernel that is up to 3.5x faster than Flash Attention v2 on NVIDIA Blackwell GPUs.
NVIDIA highlights performance per watt as the key metric determining the profitability of AI infrastructure in power-constrained environments, showcasing the overwhelming efficiency of the Blackwell platform.
How to maximize performance by optimizing GLM-5.2 with NVFP4 and TP=4 settings in an NVIDIA B200 environment
Google Cloud has launched Confidential G4 VMs equipped with NVIDIA RTX PRO 6000 Blackwell GPUs, strengthening the security of AI workloads.
AWS has launched Amazon EC2 G7 instances equipped with NVIDIA RTX PRO 4500 Blackwell GPUs, significantly boosting AI inference performance.
The NVIDIA Blackwell GB300 NVL72 platform recorded overwhelming performance in the AgentPerf benchmark, with power efficiency 20 times higher than Hopper.
Google Cloud has built a powerful security infrastructure to support Apple's Private Cloud Compute (PCC).
The real performance of AI infrastructure is determined not by FLOPS but by cost per token.