Back to home

alibaba

11 articles tagged with this topic

qwenalibaba

Alibaba's Qwen runs on two RTX 3060s — but the default config is 7× slower

Reddit user hit 400 t/s on Alibaba's Qwen3 125B with two RTX 3060s. Started at 36 t/s — an 11× gap, all from one default-config trap.

1d ago2 min read
alibabacloud-ai

Alibaba Frames AI Capex as a Three-Year Payback Story

Alibaba FY27 Q1: AI ARR RMB 49.5B (35% of external cloud). CEO Wu claims 3-year capex payback — China's first quantifiable AI ROI anchor.

Aug 212 min read
alibabaqwen

Alibaba Profit Plunges 75%: A Globally Synchronized AI Capex War Signal

Alibaba quarterly profit crashed 75%+ with $10B capex. Proof hyperscalers entering hypercapex. Builders must recalibrate Qwen, cloud pricing, GPU.

Aug 212 min read
qwenalibaba

Qwen Teases Midsize Open Model Next Week—China's Open-Source Push Accelerates

Qwen community manager teased a midsize open-source model for next week on Discord, continuing the 2025 release cadence. Community pegs 32B-72B.

Aug 192 min read
alibabaqwen

Alibaba RISC-V CPU runs 27B Qwen at 30 tps — Nvidia isn't the only bet

Alibaba's XuanTie C950 RISC-V chip ran a 27B Qwen model at 30 tps — no Nvidia required. China's domestic AI stack — chip and model — now works end-to-

Aug 182 min read
qwenalibaba

Alibaba's 27B Ties GPT-5.6 — Small Models Are Good Enough

Alibaba's Qwen 3.8 27B tied GPT-5.6 Luna at 52 on Artificial Analysis, trailing GLM-5.2 (28x larger) by one point. The "are small models enough?" deba

Aug 182 min read
alibabaQwen

Qwen3.8-27B benchmarks tie GPT-5.6 — small models can now take on big models

Artificial Analysis scored Alibaba's Qwen3.8-27B tying DeepSeek V4 and GPT-5.6 Luna Max. Enterprise private AI deployment economics may be quietly shi

Aug 172 min read
alibabaqwen

Qwen 3B Downloads: Open-Source AI's Center of Gravity Shifts East

Qwen open-weight models hit 3B downloads in 6 months, surpassing Llama and Gemma. Open-weight's center of gravity shifts east to China—the AI Android

Aug 152 min read
qwenalibaba

Qwen's 2.4T Model Makes Local AI a Luxury Few Can Afford

Qwen's new model hits 2.4T parameters; weights alone take 1.2TB. Local AI is becoming a luxury — enterprise private deployment just got redrawn.

Aug 132 min read
appleqwen

Apple Adopts Alibaba's Qwen: A Different Solution for China's AI Distribution

Apple Intelligence integrating Qwen on its China site isn't a symbol of "domestic substitution"—it's distribution being repriced within regulatory con

Aug 82 min read
deepseekalibaba

India’s AI Cost Bet Is Shifting Toward Chinese Open Models

Indian startups are using DeepSeek, Alibaba, and Moonshot open-weight models to cut AI costs by an order of magnitude, signaling a shift in global pri

Jul 182 min read