[Klaud Cold] docs: add News section + Introduction header to README (EN + ZH) (#1846)

* docs: add News section + Introduction header to README (EN + ZH)

Add a "News" section (between the language switcher and the intro paragraph)
highlighting InferenceX milestones, and wrap the existing intro paragraph in an
"Introduction" header. Mirrored in README_zh.md per the bilingual-README rule.

News items: InferenceX v1 launch (2025/10), InferenceX v2 + GB300 NVL72
(2026/02), DeepSeek V4 Day 0 (2026/04), MiniMax-M3 Day 0 (2026/06).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: reword GB300 NVL72 news item (added to InferenceX & continuously benchmarked)

Update the link text to "SGLang Maintainer Lmsys Blog" in README.md and
README_zh.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: reword DeepSeek V4 news item + add dashboard link

DeepSeek V4 Pro 1.6T: continuous benchmarks live since Day 0, with article +
dashboard (preset=dsv4-launch) links. Mirrored in README.md and README_zh.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: reword MiniMax M3 news item for consistency

MiniMax M3: continuous benchmarks live since Day 0 (dashboard). Mirrored in
README.md and README_zh.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: drop link-wrapping parens in News bullets; remove redundant article links

- News bullets now end with bare [label](link) (no surrounding parentheses).
- Remove the "Full Article Write Up for InferenceXv1/v2" lines (now covered by
  the News section).
Mirrored in README.md and README_zh.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: add [2026/03] News item for newly added models

Kimi K2.5 (Kimi 2.7-Code arch), Qwen3.5-397B, GLM5 (GLM5.1 arch), MiniMax M2.5
(MiniMax M2.7 arch). Mirrored in README.md and README_zh.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: functionstackx

README.md

@@ -11,16 +11,23 @@
 
 </div>
 
+## News
+
+- **[2026/06]** 🔥 MiniMax M3: continuous benchmarks live since Day 0 [dashboard](https://inferencex.semianalysis.com/inference?preset=minimax-m3-launch)
+- **[2026/04]** 🔥 DeepSeek V4 Pro 1.6T: continuous benchmarks live since Day 0 [article](https://newsletter.semianalysis.com/p/deepseekv4-16t-day-0-to-day-43-performance), [dashboard](https://inferencex.semianalysis.com/inference?preset=dsv4-launch)
+- **[2026/03]** Added Kimi K2.5 (same architecture as Kimi 2.7-Code), Qwen3.5-397B, GLM5 (same arch as GLM5.1), and MiniMax M2.5 (same arch as MiniMax M2.7) [dashboard](https://inferencex.semianalysis.com/)
+- **[2026/02]** GB300 NVL72: added to InferenceX & continuously benchmarked [SGLang Maintainer Lmsys Blog](https://www.lmsys.org/blog/2026-02-20-gb300-inferencex/)
+- **[2026/02]** 🔥 InferenceX v2 launch — NVIDIA Blackwell vs AMD vs Hopper [article](https://newsletter.semianalysis.com/p/inferencex-v2-nvidia-blackwell-vs)
+- **[2025/10]** 🔥 InferenceX (formerly InferenceMAX) v1 launch [article](https://newsletter.semianalysis.com/p/inferencemax-open-source-inference)
+
+## Introduction
+
 InferenceX™ (formerly InferenceMAX) is an inference performance research platform dedicated to continually analyzing & benchmarking the world’s most popular open-source inference frameworks used by major token factories and models to track real performance in real time. As these software stacks improve, InferenceX™ captures that progress in near real-time, providing a live indicator of inference performance progress. A [open sourced](https://github.com/SemiAnalysisAI/InferenceX-app) live dashboard  is available for free publicly at https://inferencex.com/. 
 
 > [!IMPORTANT]
 > Only [SemiAnalysisAI/InferenceX](https://github.com/SemiAnalysisAI/InferenceX) repo contains the Official InferenceX™ result, all other forks & repos are Unofficial. The benchmark setup & quality of machines/clouds in unofficial repos may be differ leading to subpar benchmarking. Unofficial must be explicitly labelled as Unofficial.
 > Forks may not remove this disclaimer
 
-[Full Article Write Up for InferenceXv2](https://newsletter.semianalysis.com/p/inferencex-v2-nvidia-blackwell-vs)
-[Full Article Write Up for InferenceXv1](https://newsletter.semianalysis.com/p/inferencemax-open-source-inference)
-
-
 <img width="1150" height="665" alt="image" src="https://github.com/user-attachments/assets/1e9738d4-6fb2-4cd7-a3e9-e6b2e03faed1" />
 <img width="1098" height="655" alt="image" src="https://github.com/user-attachments/assets/5b363271-69b9-4bd2-b85d-b33b9c16f50f" />
 

README_zh.md

@@ -11,16 +11,23 @@
 
 </div>
 
+## 新闻
+
+- **[2026/06]** 🔥 MiniMax M3：自 Day 0 起持续进行基准测试 [仪表盘](https://inferencex.semianalysis.com/inference?preset=minimax-m3-launch)
+- **[2026/04]** 🔥 DeepSeek V4 Pro 1.6T：自 Day 0 起持续进行基准测试 [文章](https://newsletter.semianalysis.com/p/deepseekv4-16t-day-0-to-day-43-performance)，[仪表盘](https://inferencex.semianalysis.com/inference?preset=dsv4-launch)
+- **[2026/03]** 新增 Kimi K2.5（与 Kimi 2.7-Code 架构相同）、Qwen3.5-397B、GLM5（与 GLM5.1 架构相同）、MiniMax M2.5（与 MiniMax M2.7 架构相同）[仪表盘](https://inferencex.semianalysis.com/)
+- **[2026/02]** GB300 NVL72：已加入 InferenceX 并持续进行基准测试 [SGLang 维护者 LMSYS 博客](https://www.lmsys.org/blog/2026-02-20-gb300-inferencex/)
+- **[2026/02]** 🔥 InferenceX v2 发布——NVIDIA Blackwell 对比 AMD 对比 Hopper [文章](https://newsletter.semianalysis.com/p/inferencex-v2-nvidia-blackwell-vs)
+- **[2025/10]** 🔥 InferenceX（前身为 InferenceMAX）v1 发布 [文章](https://newsletter.semianalysis.com/p/inferencemax-open-source-inference)
+
+## 简介
+
 InferenceX™（前身为 InferenceMAX）是一个推理性能研究平台，致力于持续分析与基准测试全球最受欢迎的开源推理框架——这些框架被各大 Token 工厂与模型广泛采用，以实时追踪其真实性能。随着这些软件栈不断改进，InferenceX™ 会以近乎实时的方式捕捉这些进展，提供一个反映推理性能进步的实时指标。我们在 https://inferencex.com/ 上免费公开提供了一个[开源](https://github.com/SemiAnalysisAI/InferenceX-app)的实时仪表盘。
 
 > [!IMPORTANT]
 > 只有 [SemiAnalysisAI/InferenceX](https://github.com/SemiAnalysisAI/InferenceX) 仓库才包含官方的 InferenceX™ 结果，所有其他派生（fork）与仓库均为非官方。非官方仓库的基准测试设置以及机器/云环境的质量可能存在差异，从而导致基准测试结果欠佳。非官方仓库必须明确标注为“非官方（Unofficial）”。
 > 派生仓库不得移除本免责声明。
 
-[InferenceXv2 完整文章解读](https://newsletter.semianalysis.com/p/inferencex-v2-nvidia-blackwell-vs)
-[InferenceXv1 完整文章解读](https://newsletter.semianalysis.com/p/inferencemax-open-source-inference)
-
-
 <img width="1150" height="665" alt="image" src="https://github.com/user-attachments/assets/1e9738d4-6fb2-4cd7-a3e9-e6b2e03faed1" />
 <img width="1098" height="655" alt="image" src="https://github.com/user-attachments/assets/5b363271-69b9-4bd2-b85d-b33b9c16f50f" />