Organisation · lab · CN

Alibaba / Qwen

The Qwen family of open-weight and frontier models, developed by the company's cloud division.

30 entries · April 2023 – September 2026

Alibaba is the Chinese technology and e-commerce group whose Qwen family, developed by its cloud division, has become one of the most widely used lines of open-weight AI models. Beginning with the first Qwen releases in 2023, the company has published a broad range of models — spanning language, coding, vision and reasoning — many under permissive licences, and credits them with hundreds of millions of downloads and a large ecosystem of derivative fine-tunes. That open strategy has made Chinese labs, Alibaba among them, a significant source of the freely available models the wider field depends on. By 2026 it was releasing frontier-scale systems on a roughly quarterly cadence, including trillion-parameter models such as Qwen3-Max, while — like several rivals — keeping its very largest models closed and API-only even as it open-sourced smaller ones.

Category
Chinese AI labs
Founded
1999
HQ
Hangzhou, CN
Key people
Eddie Wu

Appears alongside

Featured in threads

Tracks

  • Models & capabilities 21
  • Open weights & ecosystem 17
  • Benchmarks & progress 5
  • Compute & infrastructure 4
  • Government & policy 3
  • Safety & alignment 2
  • Security & misuse 2
  • Money & business 2
  • Culture & impact 1
  • Labs & people 1

  1. Alibaba unveils its Zhenwu V900 chip and Qwen 4 plans

    Alibaba's chip unit claimed triple the performance of its prior accelerator and clusters of up to 500,000 chips, while its Qwen team said a four-tier Qwen 4 family is training with no release date set.

    Compute & infrastructure · Models & capabilities

  2. Study detects reward hacking from models' internal representations

    Cheap difference-of-means probes on frontier open-weight models matched expensive LLM-judge monitors at catching reward hacking, at a fraction of the compute cost.

    Zhipu AI, Moonshot AISafety & alignment · Benchmarks & progress

  3. Anthropic details how states and criminals misused Claude

    Anthropic's September threat report said Claude had been used to write missile-guidance software for a Yemen weapons cell and to support biological-weapons research, alongside autonomous cyberattacks, state surveillance and the Chinese distillation campaigns.

    Anthropic, Moonshot AISecurity & misuse · Safety & alignment

  4. US agencies accuse six Chinese labs of distilling US AI models

    A joint NSA/CISA/FBI advisory named DeepSeek, Moonshot, Alibaba, MiniMax, StepFun and Z.AI as running systematic distillation campaigns against Claude, GPT, Gemini and Grok since late 2024; China's Commerce Ministry rejected it and threatened countermeasures.

    US Government, Chinese GovernmentSecurity & misuse · Government & policy

  5. Alibaba completes a record Hong Kong share placement to fund its AI buildout

    The HK$80bn (~US$10.2bn) raise is the largest-ever follow-on share offering by a Hong Kong-listed company, priced at an 8.4% discount.

    Money & business · Compute & infrastructure

  6. Alibaba releases Qwen3.8-Flash-Next, an open-weight preview of a new architecture

    A 125-billion-parameter model pairs a 51-billion-parameter phrase-lookup table held in ordinary server memory, cutting training cost to roughly a ninth of its predecessor.

    Open weights & ecosystem · Models & capabilities

  7. Thomson Reuters launches an in-house AI model built on open weights

    Thomson Reuters said the model cost about $40 million to build atop open weights rather than pretraining from scratch, and put a smaller variant on Hugging Face.

    Thomson ReutersModels & capabilities · Open weights & ecosystem

  8. Alibaba reports a 22-quarter high in AI cloud growth

    Revenue from AI Cloud and Compute Services rose 45% year on year to roughly $7.1bn, while quarterly capital spending climbed about 75% to near $10bn.

    Money & business · Compute & infrastructure

  9. Alibaba's Qwen models pass 3 billion downloads

    Hugging Face's own count, which excludes Alibaba's separate ModelScope hub, put the total closer to 2 billion — about a third below Alibaba's headline figure.

    Meta AI, Google DeepMindOpen weights & ecosystem

  10. Apple reportedly trained a China-specific AI model with Alibaba

    Apple would also offer Alibaba's Qwen as a separate option in China, Reuters reported, after Chinese regulators reportedly logged Apple's service the previous month.

    AppleGovernment & policy · Models & capabilities

  11. Alibaba open-weights its largest model, a 2.4-trillion-parameter flagship

    The mixture-of-experts model activates 95 billion of its 2.4 trillion parameters per token; the open weights are text-only, unlike the hosted version's vision input and larger context.

    Open weights & ecosystem · Models & capabilities

  12. Alibaba unveils Qwen3.8-Max, its largest model, ahead of open-weight release

    2.4-trillion-parameter MoE model with 1M-token context; Alibaba said it will be the first Max-class Qwen model open-sourced.

    Open weights & ecosystem · Models & capabilities

  13. China's AI companion law takes effect, forcing Doubao and Qwen to shut agent features

    Rather than add the anti-addiction and instant-exit features the rules required, ByteDance and Alibaba simply switched off their personalised AI-agent tools instead.

    ByteDance, Chinese GovernmentGovernment & policy · Culture & impact

  14. Zhipu (Z.ai) releases GLM-5, trained entirely on Huawei Ascend chips

    Released under the MIT licence, the 744-billion-parameter model scored 77.8% on SWE-bench Verified, days ahead of new Alibaba and ByteDance model launches.

    Zhipu AI, MiniMaxModels & capabilities · Open weights & ecosystem · Compute & infrastructure

  15. Alibaba unveils Qwen3-Max, its first trillion-parameter model

    Unlike most of Alibaba's Qwen line, the model is closed-weight and API-only, released in separate instruct and thinking modes and scoring 69.6 on SWE-bench.

    Models & capabilities · Benchmarks & progress

  16. Alibaba releases Qwen3-Next, Qwen3-VL and Qwen3-Omni

    Three architecture updates in one month: a sparse hybrid-attention base model, an updated vision-language line, and an Apache-licensed model handling text, image, audio and video.

    Open weights & ecosystem · Models & capabilities

  17. Alibaba releases Qwen3-Coder

    The mixture-of-experts model activates 35B of its 480B parameters per token and shipped under an Apache 2.0 licence with a command-line coding agent tool.

    Open weights & ecosystem · Models & capabilities

  18. Alibaba releases Qwen3 model family

    Open-weight family (dense and MoE, up to 235B-A22B) trained on 36 trillion tokens across 119 languages, Apache 2.0.

    Open weights & ecosystem · Models & capabilities

  19. Alibaba releases Qwen2.5-Omni multimodal model

    The 7B open-weight model takes text, images, audio and video as input and streams natural speech output, using a 'Thinker-Talker' architecture to separate reasoning from voice generation.

    Open weights & ecosystem · Models & capabilities

  20. ETH Zurich's 'Proof or Bluff?' finds reasoning models fail proof-based USAMO 2025

    Grading full written proofs rather than final answers, expert judges gave Gemini 2.5 Pro 24% and every other tested model under 5%, out of a possible 100%.

    Google DeepMind, OpenAIBenchmarks & progress

  21. Alibaba releases QwQ-32B (full release)

    Alibaba's Qwen team said reinforcement learning let a 32-billion-parameter model reach performance comparable to DeepSeek-R1's 671-billion-parameter model, under an Apache 2.0 licence.

    DeepSeekOpen weights & ecosystem · Models & capabilities · Benchmarks & progress

  22. 01.AI stops pre-training new large models from scratch

    Kai-Fu Lee said pretraining large models from scratch was no longer viable for a startup, and pivoted 01.AI toward fine-tuning DeepSeek and Qwen for enterprise clients.

    01.AI, DeepSeekLabs & people

  23. Alibaba releases Qwen2.5-Max

    Unlike most of Alibaba's Qwen line, Max was released as a proprietary API-only model, pretrained on over 20 trillion tokens, which Alibaba said beat DeepSeek-V3 on several benchmarks.

    Models & capabilities · Benchmarks & progress

  24. Alibaba releases Qwen2.5-VL

    Vision-language family in 3B, 7B and 72B sizes with wider OCR-language coverage and computer-control agent features, licensed differently by size.

    Open weights & ecosystem · Models & capabilities

  25. Alibaba releases QwQ-32B-Preview reasoning model

    Built on Qwen2.5-32B and released under an Apache 2.0 licence, Alibaba flagged the model could enter circular reasoning loops and mix languages mid-response.

    Open weights & ecosystem · Models & capabilities

  26. Alibaba releases Qwen2.5-Coder

    Open-weight coding-specialised model family built on Qwen2.5, aimed at competing with DeepSeek-Coder and closed coding models.

    Open weights & ecosystem · Models & capabilities

  27. Alibaba releases Qwen2.5 model family

    Alibaba's release spanned seven sizes from 0.5B to 72B parameters, plus dedicated coding and maths variants, trained on 18 trillion tokens.

    Open weights & ecosystem · Models & capabilities

  28. Alibaba releases Qwen2

    Five model sizes from 0.5B to 72B parameters, trained on 27 additional languages beyond English and Chinese, with the smaller sizes under Apache 2.0.

    Open weights & ecosystem · Models & capabilities

  29. Alibaba open-sources Qwen-7B

    The 7-billion-parameter model, pretrained on over 2.2 trillion tokens, was released alongside a chat-tuned variant and pitched against Meta's Llama on benchmark scores.

    Open weights & ecosystem · Models & capabilities

  30. Alibaba launches Tongyi Qianwen chatbot

    Launched without advance notice and restricted to corporate clients and select media on an invite-only basis; Alibaba did not disclose a parameter count.

    Models & capabilities

In the commentary

Pieces from around the web that discuss Alibaba / Qwen. External links.

Also mentioned in 35 entries

Referenced in passing — Alibaba / Qwen isn't the main subject of these.