• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
AI

Alibaba releases open weights for Qwen-Image-2.1

Alibaba’s Qwen team says the 7-billion-parameter image model supports transparent image layers and up to 10 reference images for editing.

10 Sources, 20d ago, first seen 20d ago

TLDR

Alibaba’s Qwen team announced open weights for Qwen-Image-2.1, which combines image generation and editing in one model. The team says it natively generates and edits RGBA layers—images with transparency—and offers precise local editing while preserving fidelity in portraits and products. It also claims accelerated processing of multi-image inputs.

Combined views

—

10 Sources, first seen 20d ago

— likes— comments— saves— reposts

Combined views

—

10 Sources, first seen 20d ago

— likes— comments— saves— reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

10 Sources

Kris 🫪🫪@krisolo7AI is moving too fast and 99% of your feed is engagement bait. Here is what actually happened in the last 24 hours: 1. Alibaba open-weights Qwen-Image-2.1 A unified diffusion architecture delivering native 2K image generation and editing in a single lightweight checkpoint. Open weights drop today with community quants already wiring it into ComfyUI. 2. Runway shows off interactive real-time video streaming Moving past static text-to-video batch jobs. Runway's new architecture turns generative video into a sub-second steerable live stream you control continuously in real time. 3. Anthropic postpones its IPO to November 2026 Following OpenAI, Anthropic is delaying its public listing as it eyes a $2T valuation and up to $100B in fresh private capital to fund monster cluster capex without Wall Street micromanagement. 4. Frontier AI labs slapped with antitrust lawsuit alleging illegal slowdown cartel A federal class action filed in California accuses OpenAI, Anthropic, Google, and SpaceXAI of secretly colluding on safety delays and release pacing to throttle competitive open-weights development. 5. Cua drops CUA-S1: System 1 models for computer use Routing trivial UI clicks to multi-billion parameter reasoning models is burning compute. CUA-S1 provides lightweight, reflexive visual action execution for desktop agent fleets. 6. The "Chief of Staff" pattern takes over autonomous coding fleets As single-turn CLI agents hit reasoning walls, developers are standardizing on a supervisor-worker loop: one coordinator managing state and verifying diffs while isolated Claude Code sessions execute tasks in parallel. Bookmark this to stay ahead. What was the biggest drop for you today?20d
JEFF l AI 🦞@0xjeffaiQWEN-Image-2.1 오픈웨이트 공개 사이즈는 7b 밖에 안하네요;; 로컬세팅을 준비중인데 GLM-5.3 flash ,영상은 Minimax-H3 , 이미지는 QWEN-image-2.1 로 가능하겠네요.20d
AI daily news@AIDailyNews_2x- 🎨 Alibaba releases the Qwen-Image-2.1 image generation and editing model with a 7B architecture https://qwen.ai/blog?id=qwen-image-2.1 - ⚠️ Pentagon investigation: errors and overreliance on Maven AI led to an attack on a school in Iran https://huggingnews.com/war/pentagon-probe-blames-ai-overreliance-for-strike-that-killed-123-childre-c7139471 - 📈 Anthropic plans a November IPO after reaching a $65B annual revenue run rate in July https://huggingnews.com/ai/anthropic-targets-record-november-ipo-and-new-model-to-combat-openai-ast-a41f99d820d
Turtle_studio_@turtle_studio_@anthropic의 IPO 올해 11월에 계획이라고 합니다. 엔트로픽의 자금조달. AI시대의 그리고 주식시장의 또 하나의 큰 페러다임이 될수 있는 순수 프론티어 AI기업의 상장. 이 IPO는 우리의 AI의 변화.우리의 주식잔고에 어떠한 변화를 맞이하게 될까요.?? @openai @spacex20d
WSAZ NewsChannel 3@WSAZnewsLawsuit says Anthropic, OpenAI, SpaceXAI and Google made illegal agreement on AI slowdown https://www.wsaz.com/2026/09/20/lawsuit-says-anthropic-openai-spacexai-google-made-illegal-agreement-ai-slowdown/?utm_source=twitter&utm_medium=social&utm_campaign=snd&utm_content=wsaz20d
AI议会@RSIAgent@kaero @ModelScope2022 魔搭当天就能训 Qwen-Image-2.1，分发×微调闭环比「只丢权重」更吓人。开源图模真正杀伤力：社区当晚就能长出垂直 LoRA。19d
Kyssta@kysstalolQwen-Image-2.1 (7B open weights) scored 60.28 on image evals, beating Google's nano banana 2.0 (59.82) and nano banana pro (59.45) — a 320M-param open model outperforming proprietary ones. https://cellcog.ai/blog/qwen-image-2-1/.19d
felix@felix98813Qwen-Image-2.1 开源了，号称 7B 生图加编辑、原生透明通道、2K 直出，官方分数压过一堆闭源。 但学生先别看榜，先看三笔账： 第一笔，显存。标着 7B，别以为能塞进笔记本——管线还带着视觉编码器，官方 BF16 权重三十多 GB。实测 4090 BF16 出 1024×1024 大约 21 秒、吃 30GB 出头，2K 直接奔两分钟；16GB 的卡，1024 勉强能塞，2K 基本就是墙。 第二笔，许可。开源权重不等于能拿去接单。官方模型卡写得很清楚：研究评估随便玩，商用要另拿授权。课设海报没问题，想帮人做商品图，先把许可证读完。 第三笔，防打脸。中文海报、透明贴纸、连续局部编辑、角色一致性锁不锁得住——这些比"开源榜第一"好验多了。已经有人说角色三视图还差点意思。别转发布稿了，拿同一句提示词去撞 GPT 和豆包，当场见分晓。19d
Keigo@KeigoMDQwen image 2.1 お試し 叩き台でこれが出てきて感動している19d
Gravure Lab@gravurelabQwen-Image-2.119d
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    10 Sources

    Kris 🫪🫪@krisolo7AI is moving too fast and 99% of your feed is engagement bait. Here is what actually happened in the last 24 hours: 1. Alibaba open-weights Qwen-Image-2.1 A unified diffusion architecture delivering native 2K image generation and editing in a single lightweight checkpoint. Open weights drop today with community quants already wiring it into ComfyUI. 2. Runway shows off interactive real-time video streaming Moving past static text-to-video batch jobs. Runway's new architecture turns generative video into a sub-second steerable live stream you control continuously in real time. 3. Anthropic postpones its IPO to November 2026 Following OpenAI, Anthropic is delaying its public listing as it eyes a $2T valuation and up to $100B in fresh private capital to fund monster cluster capex without Wall Street micromanagement. 4. Frontier AI labs slapped with antitrust lawsuit alleging illegal slowdown cartel A federal class action filed in California accuses OpenAI, Anthropic, Google, and SpaceXAI of secretly colluding on safety delays and release pacing to throttle competitive open-weights development. 5. Cua drops CUA-S1: System 1 models for computer use Routing trivial UI clicks to multi-billion parameter reasoning models is burning compute. CUA-S1 provides lightweight, reflexive visual action execution for desktop agent fleets. 6. The "Chief of Staff" pattern takes over autonomous coding fleets As single-turn CLI agents hit reasoning walls, developers are standardizing on a supervisor-worker loop: one coordinator managing state and verifying diffs while isolated Claude Code sessions execute tasks in parallel. Bookmark this to stay ahead. What was the biggest drop for you today?20d
    JEFF l AI 🦞@0xjeffaiQWEN-Image-2.1 오픈웨이트 공개 사이즈는 7b 밖에 안하네요;; 로컬세팅을 준비중인데 GLM-5.3 flash ,영상은 Minimax-H3 , 이미지는 QWEN-image-2.1 로 가능하겠네요.20d
    AI daily news@AIDailyNews_2x- 🎨 Alibaba releases the Qwen-Image-2.1 image generation and editing model with a 7B architecture https://qwen.ai/blog?id=qwen-image-2.1 - ⚠️ Pentagon investigation: errors and overreliance on Maven AI led to an attack on a school in Iran https://huggingnews.com/war/pentagon-probe-blames-ai-overreliance-for-strike-that-killed-123-childre-c7139471 - 📈 Anthropic plans a November IPO after reaching a $65B annual revenue run rate in July https://huggingnews.com/ai/anthropic-targets-record-november-ipo-and-new-model-to-combat-openai-ast-a41f99d820d
    Turtle_studio_@turtle_studio_@anthropic의 IPO 올해 11월에 계획이라고 합니다. 엔트로픽의 자금조달. AI시대의 그리고 주식시장의 또 하나의 큰 페러다임이 될수 있는 순수 프론티어 AI기업의 상장. 이 IPO는 우리의 AI의 변화.우리의 주식잔고에 어떠한 변화를 맞이하게 될까요.?? @openai @spacex20d
    WSAZ NewsChannel 3@WSAZnewsLawsuit says Anthropic, OpenAI, SpaceXAI and Google made illegal agreement on AI slowdown https://www.wsaz.com/2026/09/20/lawsuit-says-anthropic-openai-spacexai-google-made-illegal-agreement-ai-slowdown/?utm_source=twitter&utm_medium=social&utm_campaign=snd&utm_content=wsaz20d
    AI议会@RSIAgent@kaero @ModelScope2022 魔搭当天就能训 Qwen-Image-2.1，分发×微调闭环比「只丢权重」更吓人。开源图模真正杀伤力：社区当晚就能长出垂直 LoRA。19d
    Kyssta@kysstalolQwen-Image-2.1 (7B open weights) scored 60.28 on image evals, beating Google's nano banana 2.0 (59.82) and nano banana pro (59.45) — a 320M-param open model outperforming proprietary ones. https://cellcog.ai/blog/qwen-image-2-1/.19d
    felix@felix98813Qwen-Image-2.1 开源了，号称 7B 生图加编辑、原生透明通道、2K 直出，官方分数压过一堆闭源。 但学生先别看榜，先看三笔账： 第一笔，显存。标着 7B，别以为能塞进笔记本——管线还带着视觉编码器，官方 BF16 权重三十多 GB。实测 4090 BF16 出 1024×1024 大约 21 秒、吃 30GB 出头，2K 直接奔两分钟；16GB 的卡，1024 勉强能塞，2K 基本就是墙。 第二笔，许可。开源权重不等于能拿去接单。官方模型卡写得很清楚：研究评估随便玩，商用要另拿授权。课设海报没问题，想帮人做商品图，先把许可证读完。 第三笔，防打脸。中文海报、透明贴纸、连续局部编辑、角色一致性锁不锁得住——这些比"开源榜第一"好验多了。已经有人说角色三视图还差点意思。别转发布稿了，拿同一句提示词去撞 GPT 和豆包，当场见分晓。19d
    Keigo@KeigoMDQwen image 2.1 お試し 叩き台でこれが出てきて感動している19d
    Gravure Lab@gravurelabQwen-Image-2.119d
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet