Alibaba rolls out Qwen4-Max with cheaper APIs and a million-token context
Alibaba's flagship Qwen4-Max is now generally available on QwenCloud, ending a limited preview that started November 11. The 2.4-trillion-parameter model reads images, handles up to a million tokens of context, and Alibaba says it cut average task costs on its own benchmark by roughly 58%.
- Original source: technode.com.
- The source published it on Nov 16, 2026 at 6:35 AM.
- No second outlet has reported it yet. Treat it as a single-source claim.
Alibaba moved Qwen4-Max into general availability on Monday, opening its flagship model to all users through QwenCloud after a limited preview that began on November 11. The model also shows up in Qwen Studio, QwenWork and Qoder. It runs on a mixture-of-experts architecture with 2.4 trillion parameters.
The context window stretches to one million tokens, and the model takes images directly rather than through a separate pipeline. Alibaba says it burns fewer tokens than Qwen3.8-Max on comparable tasks, which helps the price story land.
Cheaper API pricing and the leaner token use pushed the average task cost on Alibaba's own Work360 benchmark down to RMB 11.16, from RMB 26.58 — a reduction of about 58%.
Alongside the model, Alibaba shipped a plugin that lets agents hand over an initial version of their work and then revise it as users give feedback. The company also said smaller Qwen4 models are headed for open-source release.
TopicsAlibaba · Qwen4-Max · QwenCloud · Qwen3.8-Max · Qwen Studio · QwenWork · Qoder · Qwen4
Source: technode.com
Written by AI from verified reporting.
More stories
- Tech & AI ft.com 0 viewsOhio voters push back on Amazon's $17B AI data centre plan
- Tech & AI x.com 0 viewsAnthropic lines up Russell 1000 and Nasdaq 100 entries after IPO
- Tech & AI x.com 1 viewUS details Megaspeed probe, pushes back on 'inaccurate' reports
- Tech & AI blockchain.news 1 viewGrid Robotics paper pitches 'agentic robotics' powered by GPT-6 Astra

