Best English Newsletters for China AI News

Quick Routing: What You Want vs. Where to Get It

No single newsletter covers every dimension of China AI. Use this table to match your need to the right source — and understand what each source is not good for.

Top English Newsletters and Sites for China AI News (2026)

1. RadarAI Weekly English Digest

Focus: Builder-facing signal — model releases, open-source drops, API changes, and benchmark comparisons from Chinese AI labs.
Cadence: Weekly (Sunday)
Best for: Builders integrating Chinese models (Qwen3, DeepSeek-R1, Kimi k2) into products, engineers tracking API pricing and capability changes, PMs deciding when to switch base models.
Not good for: Geopolitical analysis, investor-grade funding data, China tech policy context.

In Q2 2026, RadarAI covered both Qwen3’s Apache 2.0 release (April 2026, 235B flagship MMLU 87.1, 30B-A3B MoE variant with only 3B active params) and DeepSeek-R1-0528 (May 2026, AIME 72.6%) in the same digest week they published, with direct links to the HuggingFace model cards and GitHub repos. Most general AI newsletters covered these 4-7 days later, often without the benchmark tables.

→ View RadarAI China AI updates feed

2. ChinAI (Jeffrey Ding, Substack)

Focus: Translated analysis from Chinese primary sources — academic papers, government documents, industry reports, and policy discussions.
Cadence: Weekly
Cost: Free
Best for: Researchers, policy analysts, people who want to understand the why behind Chinese AI strategy, not just the what.
Not good for: Same-week model release tracking, technical benchmarks, developer API access guides.

ChinAI has been running since 2018 and remains the gold standard for translated primary-source analysis. Jeffrey Ding is now at George Washington University’s Elliott School. The newsletter doesn’t try to be a model tracker — it’s for understanding the context that explains why DeepSeek open-sourced weights or why Alibaba chose Apache 2.0 for Qwen3.

3. 36Kr Global

Focus: Chinese tech and AI business news in English — funding rounds, startup profiles, enterprise AI adoption, regulatory context.
Cadence: Daily articles, no fixed newsletter
Cost: Free (website + newsletter signup)
Best for: Investors, business development, corporate strategy teams monitoring the Chinese AI industry from a commercial angle.
Not good for: Open-source technical releases, model benchmark comparisons, developer tooling updates.

36Kr Global translates and extends coverage from 36Kr, the leading Chinese tech media outlet comparable to TechCrunch in the US market. For funding rounds (including the wave of Chinese AI startups that raised in early 2026), 36Kr Global is consistently 24-48 hours ahead of Western business press.

4. BestBlogs.dev

Focus: Aggregated Chinese technical blog posts with AI-generated English summaries — from Chinese developers, researchers, and AI lab blogs.
Cadence: Daily updates
Cost: Free
Best for: Developers who want to read what Chinese engineers are writing about (architecture decisions, implementation notes, open-source project updates) without reading Chinese.
Not good for: Business/investment context, breaking news, official lab announcements (those are better tracked via GitHub/HuggingFace directly).

5. HuggingFace Daily Papers + Model Hub

Focus: Real-time academic papers and model releases — many Chinese AI labs post directly to HuggingFace.
Cadence: Real-time
Cost: Free
Best for: Technical depth, model weights download, benchmark tables, release notes.
Not good for: Business context, policy, or news that doesn’t come with a model card.

For Chinese model releases specifically: deepseek-ai on HuggingFace, Qwen on HuggingFace, and THUDM (Zhipu AI parent) are the authoritative primary sources — every newsletter ultimately links back to these.

Newsletter Comparison: Coverage Matrix

Recommended Newsletter Stacks by Use Case

Builder / Developer Stack (2 sources)

1. RadarAI weekly digest — weekly model + tool signal with benchmark data
2. HuggingFace Daily Papers — real-time model cards and weights when you need depth

Add BestBlogs.dev if you want to read what Chinese engineers are writing about implementation details.

Investor / Analyst Stack (3 sources)

1. 36Kr Global — funding rounds and business news
2. ChinAI — policy and strategic context
3. RadarAI weekly — technical signal to understand what the funding is actually building

Researcher / Policy Stack (2 sources)

1. ChinAI — primary source translation and analysis
2. CSET Georgetown publications — data-driven policy research on China AI

Supplement with SCMP Tech and MIT Technology Review for broader geopolitical framing.

What No Single Newsletter Covers (and How to Fill the Gap)

Even with the best stack, there are systematic gaps in English-language China AI coverage:

• API pricing changes in real time — Chinese AI APIs (Qwen API, DeepSeek API, Kimi API) change pricing with minimal notice. No newsletter tracks this systematically; check the official API pages directly.
• Model deprecation timelines — When Qwen2.5 gets deprecated after Qwen3 launches, the deprecation notice is typically in Chinese first. Subscribe to the GitHub release feeds directly.
• Export control implications — US export control rules affecting Chinese AI chips and models are updated without newsletter coverage. CSET and Lawfare are the best English-language sources for this.
• Developer community discussions — The real-time developer conversation happens on Chinese platforms (Zhihu, WeChat groups, Bilibili). BestBlogs.dev captures some of this but not in real time.

Frequently Asked Questions

What is the best English newsletter for China AI news?

For technical signal (model releases, benchmarks): RadarAI weekly. For strategic/policy analysis: ChinAI by Jeffrey Ding. For business and funding: 36Kr Global. No single newsletter covers all three — use a 2-source stack matched to your use case.

Is there an English newsletter that covers DeepSeek and Qwen model releases?

RadarAI covers these weekly with benchmark data. HuggingFace model hub and the QwenLM GitHub are the real-time primary sources. DeepSeek-R1-0528 (May 2026) posted benchmarks — AIME 72.6%, MATH-500 97.3% — directly to HuggingFace; newsletters that covered it with numbers were pulling from there.

What China AI newsletter is best for investors?

36Kr Global for deal flow and business context. KR-Asia as a backup. Neither covers technical benchmarks in depth — pair with RadarAI for the technical dimension of what you’re investing in.

Are there free China AI newsletters in English?

Yes. All five sources in the comparison table above offer free tiers: RadarAI, ChinAI, 36Kr Global, BestBlogs.dev, and HuggingFace. The paid options (The Information, some SCMP content) add depth but the core signal is available free.

How do I track Chinese AI news in real time, not just weekly?

For real-time: watch the GitHub release feeds for QwenLM, DeepSeek, and THUDM; watch HuggingFace model pages for new uploads; follow @deepseek_ai and @Qwen_LM on X. Newsletters are curated weekly summaries — for same-hour coverage, primary GitHub/HuggingFace feeds are the only reliable path.

What’s the difference between China AI newsletters and general AI newsletters?

General AI newsletters (The Rundown, AI Breakfast, TLDR AI) cover Western labs (OpenAI, Anthropic, Google) with depth and cover Chinese labs as occasional newsworthy items. They typically miss open-source releases from Chinese labs, miss benchmark comparisons, and have a 3-7 day lag on Chinese developments. China-specific sources cover the Chinese AI ecosystem as the primary beat, not a secondary one.

Related Pages