corpus.blog
42,697 blogs · 5,350,595 posts Claim your blog
Back to inferencex.semianalysis.com
Blog · corpus.blog/blogs/inferencex.semianalysis.com/posts

inferencex.semianalysis.com

inferencex.semianalysis.com

2026

How GLM5.3 Sparse Attention Affects HBM Memory Usageoriginal ↗
28 Sept 2026
Engrams Embedding Entendre: Codesign for Efficient DRAM/SSD Offloadingoriginal ↗
18 Sept 2026
Rubin NVL72 Agentic Inference: 67x better Performance per Dollaroriginal ↗
14 Sept 2026
TPU Inference Externalization Full Steam Aheadoriginal ↗
7 Sept 2026
DeepSeek V4 Pro on AgentX: B200 vs B300 and the KV Cache Working Setoriginal ↗
25 Aug 2026
DeepSeek V4 Pro on AgentX: GB200 vs GB300 Rack-Scale Disaggregationoriginal ↗
25 Aug 2026
DeepSeek V4 Pro on AgentX: MI355X vs B200, and the August 21 Fliporiginal ↗
25 Aug 2026
GLM 5.3 on AgentX: MI355X ATOM Beats GB300 NVL72 on Part of the Curveoriginal ↗
25 Aug 2026
GLM 5.3 on AgentX: NVIDIA Is Up to 5x Cheaper per Token at 150 tok/s/useroriginal ↗
25 Aug 2026
Kimi K3 on AgentX: MI355X ATOM Beats GB300 NVL72 on Part of the Curveoriginal ↗
25 Aug 2026
MiniMax M3 on AgentX: Why B200 and B300 Beat Their Rack-Scale GB200 NVL72 & GB300 NVL72 Counterpartsoriginal ↗
25 Aug 2026
MiniMax M3 on AgentX: B300 TRT-LLM TP2 Owns the Crownoriginal ↗
25 Aug 2026
OpenAI Jalapeño: Better Than Nvidia Blackwelloriginal ↗
25 Aug 2026
Qwen3.5 397B on AgentX: B300 FP4 Delivers 12x the Performance per Dollar of H100original ↗
25 Aug 2026
MI355X versus GB300 NVL72 Inference Performance: 20x Gap on Qwen3.5 SGLangoriginal ↗
25 Aug 2026
AgentX - InferenceXv3: Does the CUDA Moat Hold Up in Agentic Inferencing?original ↗
24 Aug 2026
Agentic Benchmark for LLM Inference: Metrics and Methodologyoriginal ↗
19 Aug 2026
A Brief Overview of Agentic Workloadsoriginal ↗
19 Aug 2026
Ultra-High Interactivity on NVIDIA GPUs? TileRT on InferenceXoriginal ↗
10 Aug 2026
Kimi K3: The Manos, The Mythos, The Legendosoriginal ↗
3 Aug 2026
Vera Rubin NVL72 vs GB200 NVL72? Inference TCO & Architecture Analysisoriginal ↗
23 Jul 2026
DeepSeekV4 1.6T Day 0 to Day 43 Performance Over Time — Huawei, GB300 NVL72, MI355X, B200original ↗
9 Jun 2026
GB300 NVL72 vs GB200 NVL72 Inference Performance & Perf per Dollar - on DeepSeek-V4-Pro 1.6T: Up to 2.83x Throughputoriginal ↗
27 May 2026
B200 NVFP4 vs H200 FP8 on GLM-5: Up to 3.65x Better Performance per Dollar with SGLang MTPoriginal ↗
26 May 2026
B200 NVFP4 vs H100 FP8 on MiniMax-M2.5: Up to 8.2x Better Performance per Dollar with vLLMoriginal ↗
26 May 2026
B200 NVFP4 vs H200 INT4 on Kimi K2.5/K2.6: Up to 2.95x Better Performance per Dollaroriginal ↗
26 May 2026
MI355X DeepSeek-V4-Pro on SGLang: 110.5x Throughput per GPU in 26 Daysoriginal ↗
26 May 2026
AMD MI355X GLM-5 Inference: Up to 40% Cheaper per Million Tokens than B200 on SGLang FP8original ↗
25 May 2026
AMD MI355X Qwen3.5 397B-A17B Inference: Up to 19x Throughput per GPU in 3 Months on SGLang FP8original ↗
25 May 2026
GB200 NVL72 vs B200 on DeepSeek R1 670B: Up to 4.4x Throughput per GPU at 125 tok/s/useroriginal ↗
23 May 2026
SGLang 0.5.6 on B200 DeepSeek R1 FP4: Up to 1.8x at Low Concurrencyoriginal ↗
2 May 2026
GB200 NVL72 vs B200 on Kimi K2.5: 3.1x from Wide EP vLLMoriginal ↗
23 Apr 2026
AMD MI355X Kimi K2.5 Inference: 7.7x Throughput, Up To 15x Interactivity in 25 Days on vLLMoriginal ↗
22 Apr 2026
InferenceX v2: NVIDIA Blackwell Vs AMD vs Hopper - Formerly InferenceMAXoriginal ↗
16 Feb 2026

2025

InferenceMAX: Open Source Inference Benchmarkingoriginal ↗
9 Oct 2025