#b200

The latest AI and developer news about #b200, with the original source and a short summary.

LlamaIndex Achieves 10x Inference Speedup for OCRia-1 with Custom Optimization Stack

The team combined mixed-precision quantization, DFlash2 speculative decoding, and custom CUDA kernels to reduce latency and improve throughput on a single GPU.

·2026.11.04 00:00

Interactive Calculator Estimates GPU Needs for Llama 3.3 70B and DeepSeek V4 Models

A new interactive tool estimates GPU requirements for serving large language models like Llama 3.3 70B and DeepSeek V4 Pro, using benchmark data for H100 and B200 hardware.

·2·2026.10.07 03:00

SemiAnalysis TPUv7 Ironwood Analysis: Costs Approximately 19% Lower Than B200 at 100 Tokens Per Second Per User

SemiAnalysis analyzed that under external buyer TCO and 100 tokens per second per user conditions for Qwen3.5 397B FP8 inference, Ironwood's cost per token is approximately 19% lower than B200.

·2·2026.09.07 09:00

If You Need 4 GPUs, GPU-Hours Are No Longer a Commodity

In the GPU rental market, the supply of multiple identical GPU clusters actually needed for real workloads is far more limited than single-GPU supply, and the pricing structure differs as well.

·2·2026.07.24 09:00

Lighthouse Attention Technique

Selective hierarchical attention was about 17x faster than standard attention at 512K context.

·2·2026.05.13 00:00

GPU Spot Prices Surge 114% in Six Weeks

NVIDIA B200 GPU spot prices surged 114% in six weeks.

·1·2026.04.28 09:00