NVIDIA
NVIDIA is an American semiconductor and technology company primarily designing and selling graphics processing units (GPUs), system-on-chip processors, and related software platforms. Its products are widely used in gaming graphics rendering, data center computing, artificial intelligence training and inference, autonomous driving, and professional visualization. NVIDIA also developed the CUDA parallel computing platform, enabling GPUs for general-purpose scientific computation, and is a principal supplier of modern AI infrastructure.
Core Facts
NVIDIA Nemotron 3.5 Super VL is a 120B-parameter MoE model with 12B active parameters, supporting up to a 1 million token context, aimed at document understanding, computer use and agent scenarios.
TensorRT is a high-performance inference optimization engine provided by NVIDIA that compiles PyTorch/ONNX models into GPU-optimized execution engines, typically improving inference speed by 2–5×.
The NVIDIA A100's HBM2e capacity is 80GB, the H100 has 80GB HBM3, and the H200 has 141GB.
Timeline (last 90 days)
NVIDIA's CentML team achieved 423 tokens per second per user on the 2.8-trillion-parameter Kimi K3 large model, a 1.68x speedup on the same architecture, thanks to a decoding kernel written in CUDA DSL and Mega Kernel Fusion technology, with about 12,000 lines of optimized code.
NVIDIA tested with 1,061 pages of production documentation; Claude found all 11 issues found by humans and flagged 7 additional issues humans had not found, cutting manual review time from more than 25 hours to less than 3 hours, an 89% reduction.
NVIDIA released the Nemotron 3.5 Transcribe model, which uses a MoE architecture with 30B total parameters and activates only 4B parameters at runtime.
The U.S. Department of Justice has opened an antitrust investigation into NVIDIA's investment partnership with Hugging Face.
The Singapore Meetup brings together engineers from inferact, NVIDIA, Alibaba Cloud, and the vLLM-Omni community, with an agenda covering directions such as high-throughput prefill, reinforcement learning, the Blackwell architecture, the Vera Rubin platform, Agent infrastructure, and interactive world models.
In NVIDIA's Ling-3.0-flash model, each token selects 8 of 512 routed experts in each MoE layer, plus one shared expert.
NeMo Speech is NVIDIA's speech toolchain, covering the training and inference pipelines for tasks such as speech recognition and speaker diarization.
NVIDIA recently released the Nemotron 3.5 Transcribe speech transcription model.
Nemotron 3.5 Transcribe's day-0 serving strategy essentially treats ecosystem collaboration as part of the product capability.
Graphics card detection relies on NVIDIA's NVML interface to read VRAM capacity and compute capability level, based on which the resolution limit and whether to enable half precision (fp16) or VRAM optimization strategies are determined.
40 more timeline events
All Facts (20)
NVIDIA Nemotron 3.5 Super VL is a 120B-parameter MoE model with 12B active parameters, supporting up to a 1 million token context, aimed at document understanding, computer use and agent scenarios.
90%VerifiedTensorRT is a high-performance inference optimization engine provided by NVIDIA that compiles PyTorch/ONNX models into GPU-optimized execution engines, typically improving inference speed by 2–5×.
90%VerifiedThe NVIDIA A100's HBM2e capacity is 80GB, the H100 has 80GB HBM3, and the H200 has 141GB.
90%VerifiedThe HBM3 memory used by NVIDIA's H100 has a bandwidth of up to 3.35 TB/s.
90%VerifiedIn the H100 NVLink architecture, 8 GPUs within a single server can be fully interconnected via NVSwitch, achieving a total bandwidth of 900 GB/s, an order-of-magnitude improvement over PCIe 5.0's 64 GB/s.
90%VerifiedNVFP4 is a 4-bit floating-point (FP4) quantization format introduced by NVIDIA, designed specifically for the hardware acceleration capabilities of Blackwell architecture GPUs
90%VerifiedA single NVIDIA A100 card has 80GB HBM2e memory and 312 TFLOPS of FP16 computing power
90%VerifiedSince 2022, the United States has imposed advanced semiconductor export controls on China, restricting the export of high-end AI training chips such as the NVIDIA A100/H100 and their successor products.
90%VerifiedNVIDIA CUDA was launched in 2006 and was originally a programming framework for enabling developers to use GPUs for general-purpose computing.
90%VerifiedThe NVIDIA H100 single-card FP64 double-precision floating-point performance reaches approximately 34 TFLOPS.
90%VerifiedxAI built a supercomputing cluster called Colossus, reportedly equipped with more than 100,000 NVIDIA H100 GPUs.
90%VerifiedNVIDIA NIM (Inference Microservices) is an inference microservice platform launched by NVIDIA that packages optimized AI models as standardized container images, supporting one-click deployment on enterprise on-premises GPUs.
90%VerifiedTraining a model at the level of GPT-4 requires tens of thousands of A100/H100 GPUs working in concert for several months.
90%VerifiedThe retail price of a single NVIDIA H100/H200 GPU is as high as $30,000–$40,000.
90%VerifiedCUDA (Compute Unified Device Architecture) is a parallel computing platform launched by NVIDIA.
90%VerifiedNVIDIA's H100/H200 series GPUs hold more than 80% of the AI training and inference market.
90%VerifiedROCm is AMD's open-source software platform for high-performance computing and AI workloads, including the HIP programming interface, which allows CUDA code to be ported to AMD GPUs at relatively low cost.
90%VerifiedNVIDIA's CUDA ecosystem is the de facto standard in the AI inference industry.
85%VerifiedQwen3.8-Flash-Next adopts a Mixture of Experts (MoE) architecture with 125B total parameters and 6B activated parameters.
85%VerifiedNVIDIA's TensorRT-LLM deeply optimizes inference speed for its hardware architecture.
85%