# Alibaba launches 2.4-trillion-parameter Qwen3.8-Max as DeepSeek releases ultra-low-cost V4-Flash, each betting on a different way to beat US AI labs
> China's Alibaba unveiled Qwen3.8-Max, its largest AI model at 2.4 trillion parameters, on August 3, sending its shares higher; simultaneously DeepSeek released V4-Flash, priced at more than 100 times below comparable Anthropic models according to analyst research; the two launches on the same day represent opposite strategies among Chinese AI firms: scale on capability versus slashing inference costs

**Meta:** type: event · date: 2026-08-03 · heads: 長期戦, 静かな変化 · 7 takes · 4 lenses · 6 regions

## Summary

[Alibaba](/ja/n/alibaba-qwen-dossier) launched Qwen3.8-Max on August 3, its largest AI model at 2.4 trillion parameters, placing it at the scale frontier alongside China's Moonshot and narrowing the gap to the largest US-developed models. On the same day, [DeepSeek](/ja/n/deepseek-dossier) released V4-Flash, described by analyst research as priced at more than 100 times below Anthropic's comparable models. The two launches represent diverging bets: Alibaba chasing raw capability, [DeepSeek](/ja/n/deepseek-v4-open-weights-2026) driving inference costs toward levels that change the economics of AI deployment globally. Israeli tech outlet CTech and Reuters both noted the dual announcement as the sharpest signal yet of [China's](/ja/entity/china) intent to compete on both dimensions simultaneously.

## Why it matters

DeepSeek's pricing signal, more than Alibaba's scale, reshapes what operators worldwide will pay to run frontier AI. A sustained 100x cost gap alters data center economics and undercuts the revenue models of US labs that have relied on premium pricing.

## What to watch

- Independent benchmarks placing Qwen3.8-Max against GPT-5 and Google Gemini Ultra on coding and reasoning.
- DeepSeek V4-Flash access terms and whether the pricing holds at scale.
- US Commerce Department response to parameter-scale advances by Chinese firms already on the entity list.

## Regional takes (batched by bias / lens)

### Wire service report from Beijing bureau, Eduardo Baptista byline; first Western wire to cover both launches together
- **Reuters** (Global, en) — Breaking report from Beijing: Alibaba launched its largest and most capable AI model yet, sending shares higher; a research firm cited DeepSeek's new product as offering cut-throat pricing more than 100 times below comparable models. First piece to juxtapose both launches as twin signals from China's AI sector.
  > "China's Alibaba on Monday unveiled its largest and most capable AI model to date, sending its shares surging, while a research firm said DeepSeek's latest product offers cut-throat pricing that is more than 100 times lower."
  Source: https://wtvbam.com/2026/08/02/alibaba-unveils-its-most-capable-ai-model-to-date-not-far-behind-moonshots-in-size/

### Israeli tech media, frames both launches as China's most direct challenge yet to US frontier AI labs
- **Calcalist CTech** (Israel, en) — Israeli tech publication positions the dual launch as a strategic split: Alibaba bets on scale with 2.4 trillion parameters, DeepSeek focuses on dramatically lower costs. Both moves are read as an intensification of China's competition with US AI leaders including OpenAI and Google.
  > "Alibaba is betting on scale with a 2.4-trillion-parameter model, while DeepSeek is focusing on dramatically lower costs as Chinese companies intensify competition with U.S. AI leaders."
  Source: https://www.calcalistech.com/ctechnews/article/rjfsc3phfl

### Technology analysis, highlights the 100x pricing gap and places both models in global competitive context
- **Tech Startups** (Global, en) — Detailed breakdown of the two models: Qwen3.8-Max at 2.4 trillion parameters pushes Alibaba into the same scale bracket as Moonshot's Kimi; DeepSeek V4-Flash is priced at more than 100 times below Anthropic's comparable offerings, per analyst research cited. Frames both as part of China's deliberate strategy to lead on both axes of the frontier.
  > "China's AI race just produced two numbers that are hard to ignore: 2.4 trillion parameters and a price gap of more than 100X."
  Source: https://techstartups.com/2026/08/03/alibaba-unveils-2-4-trillion-parameter-qwen3-8-max-as-deepseeks-new-ai-model-undercuts-anthropic-by-100x/

### unlabelled
- **Influencer Magazine** (United Kingdom, en) — 
  Source: https://influencermagazine.uk/2026/08/chinas-ai-giants-alibaba-and-deepseek-redefine-industry-standards-with-record-breaking-model-launches/
- **American Bazaar** (United States, en) — 
  Source: https://americanbazaaronline.com/2026/08/03/alibaba-unveils-its-largest-most-capable-ai-model-485698/
- **Express Tribune** (Pakistan, en) — 
  Source: https://tribune.com.pk/story/2621839/alibaba-unveils-its-largest-ai-model-yet-deepseeks-latest-model-is-ultra-low-cost
- **InfotechLead** (India, en) — 
  Source: https://infotechlead.com/artificial-intelligence/alibaba-launches-2-4-trillion-parameter-qwen3-8-max-ai-model-challenging-moonshot-and-openai-97480

## Across the graph
- Related: [[alibaba-qwen-dossier]], [[deepseek-dossier]], [[deepseek-v4-open-weights-2026]], [[qwen-37-agents-open-vs-closed-2026]]
- Entities: Corporate:alibaba Qwen, Deepseek, China

---
Canonical: https://rbtfl.xyz/ja/n/china-ai-models-0803