@ant_oss, Open Source at AntGroup; GB of @openinfradev + GB & TAB of @CloudNativeFdn; ex-Arch committee of @katacontainers; (try @ZenMuxAI ); Tweets are my own

39.963224,116.430714
Joined August 2007
Announcing Ant Group as Gold Member of Agentic AI Foundation @AgenticAIFdn. @AntGroup has always been committed to leveraging cutting-edge technology to make everyday services—from finance and payments to healthcare—accessible to all. This vision drives our AI Lab @TheInclusionAI, with the mission of "AI built by everyone, for everyone." As AI agents bring frontier LLM capabilities into real-world applications, we believe open community consensus is essential to ensuring they reliably serve everyone. Joining the Agentic AI Foundation enables us to collaborate more closely with the global community to make this a reality.
The Agentic AI Foundation (AAIF) continues to expand its global community, welcoming 305 new members across financial services, philanthropy, enterprise infrastructure, security, and technology. Announced at AGNTCon + MCPCon North America, Gates Foundation, MongoDB, Nutanix, Royal Bank of Canada, and Vultr join as Gold Members, alongside new Silver and Associate Members. Together, these organizations will help advance open standards for secure, interoperable, and trusted agentic AI, supporting adoption across industries. Read the full announcement: bit.ly/3W0K8D7
There's a new version of this post
1
5
156
The Agentic AI Foundation (AAIF) continues to expand its global community, welcoming 305 new members across financial services, philanthropy, enterprise infrastructure, security, and technology. Announced at AGNTCon + MCPCon North America, Vercel, Gates Foundation, ANT Group, MongoDB, Nutanix, Royal Bank of Canada, and Vultr join as Gold Members, alongside new Silver and Associate Members. Together, these organizations will help advance open standards for secure, interoperable, and trusted agentic AI, supporting adoption across industries. Read the full announcement: bit.ly/3W0K8D7
1
4
22
1,389
So happy we @ant_oss officially joined AAIF today. In the Agentic era, the core value of an open-source community isn't about code. It’s about who you're building with, and the Agent consensus you shape together.
4 new projects, 305 members, 15 industry sectors. Mazin Gilbert, Executive Director of AAIF, is live at AGNTCon + MCPCon NA talking growth and what’s next for the agentic AI ecosystem. Mazin is live right now. Don’t miss it. Join the livestream: bit.ly/3U6tMrS
2
5
233
Known in the community as the Rust Wizard, Ya'nan Li earned the Kata Containers 4.0 Contributor Award for his technical precision and collaborative spirit. Learn about his journey and advice for open source contributors. katacontainers.io/blog/kata-… #KataContainers
3
1
3
241
gnawux retweeted
Still using the same AI model for every task? Writing, research, coding, long docs — different tasks call for different models. Meet AI Workmates on ZenMux: GLM, Qwen, MiniMax, Kimi, and more — now 10–30% OFF 👉 zenmux.ai/ai-workmates Different models. One API. All on ZenMux. 💼 #AI #AITools #AIWorkmates
5
1
13
1,320
Huyu and I did the sandbox demo based on Kata Containers 4.0, and my blog on the demo project has been published on @katacontainers official blog: katacontainers.io/blog/kata-… . Thank you @IldikoVancsa
🤖 Made with AI
1
7
143
Back to ApacheAsia! While I now represent @ant_oss on the GB of @openinfradev and @CloudNativeFdn , my very first community contribution was right here @TheASF HADOOP-4742. I am forever grateful to ASF, Hadoop, and Hairong, the committer helped me then. #CommunityOverCode
1
152
My talk about open source agent sandbox stack, @katacontainers , @dragonfly_oss , etc on #KubeConJapan – at Yokohama-shi Naka, Kanagawa
At #KubeConJapan today, I have a topic to showcase building an agent sandbox system upon Kata 4.0 and a whole open source stack.
1
3
4
459
Pic. 4 adjusted by some model
🤖 Made with AI
1
70
gnawux retweeted
⏳ 3 days left for ZenMux Token Economics. Until July 31, 10+ popular models at DeepSeek-level pricing — up to 80% off. Same platform. Same API. Same price. Run your real tasks and see which model delivers the most value. No benchmark hype — your usage decides. Make your final tokens count 👇 zenmux.ai/token-economics #ZenMux #TokenEconomics
3
2
9
1,713
🚀 Introducing @Alipay PIBench — the first benchmark to evaluate Coding Agents on real-world payment integration. For developers & model teams: ✅ Test how well your agent handles real payments ✅ 9 products × 9 real repos × 459 expert rubrics ✅ Open-source & reproducible We are looking forward for your feedbacks. Let's build more capable payment agents together! 👉github.com/inclusionAI/PIBen… For more details, check our tech report: arxiv.org/abs/2607.14573 #PaymentSecurity #AgenticPayment #AIAgents #inclusionAI
3
5
1
20
6,092
At #KubeConJapan today, I have a topic to showcase building an agent sandbox system upon Kata 4.0 and a whole open source stack. – at Yokohama-shi Naka, Kanagawa
"Kata Containers 4.0 is less a new feature release than a platform reset to Rust." @sjvn covers what the move to Rust means for the project and why he expects Kata to become a much more popular platform going forward. opensourcewatch.beehiiv.com/…
1
4
639
gnawux retweeted
Claude Opus 5 is now live on ZenMux 🧠 Near-frontier intelligence at half the cost of Fable 5. Built for complex coding, deep research, and long-running agentic workflows—Opus 5 is our most thoughtful, proactive, and capable default model yet. 🌟 Key Highlights: 💻 New SOTA in Coding & Problem-Solving: Dominates Frontier-Bench and CursorBench; scores 3x higher than the next-best model on ARC-AGI 3. 🤖 Proactive Agency: Excels at self-verification, writing custom vision pipelines, and fixing root-cause bugs autonomously. 🔬 Scientific Breakthroughs: Significant gains in life sciences, organic chemistry, and bioinformatics. 🛡️ Our Most Aligned Model: Features the lowest deceptive behavior and highest safety adherence to date. ⚡ Same Price, More Speed: Kept at Opus 4.8 pricing ($5/M input | $25/M output). Fast mode is also available at 2.5x speed! Turn on Auto Top-Up to get 20% extra on your next 3 top-ups — up to $100 each. Top up, run Opus 5, and keep your workflow moving. 👉 zenmux.ai/anthropic/claude-o…
7
1
20
2,029
🚀 Meet LLaDA2.2-flash — the first agent-oriented MoE Diffusion LLM. Levenshtein editing—deletion, insertion, and substitution—enables self-correction during decoding. A more elegant solution to model collapse than forcing diffusion back into left-to-order generation. Performance: ⚡ 703.82 TPS on BFCL-V4 function calling 519.0 TPS on SWE-bench Verified coding ⚡ 1.64x BF16 throughput vs autoregressive models Built for high-throughput, low-latency AI agents. 🔗 GitHub: github.com/inclusionAI/LLaDA… 🤗 Hugging Face: huggingface.co/inclusionAI/L… 📄 Technical Report:github.com/inclusionAI/LLaDA… #LLaDA #OpenSource #DiffusionLLM #AIAgents
177
154
157
391
674,548
💡Meet LLaDA2.2-flash — world-first large-scale agentic diffusion model. We didn’t solve model collapse by making dLLMs autoregressive again. We gave them real editing primitives: DELETE, INSERT, and SUBSTITUTE. Native structural self-correction during parallel decoding. No strings attached. ⚡ 703.82 TPS on BFCL-V4 ⚡ 519.0 TPS on SWE-bench Verified ⚡ Native 128K context via Block Routing Check the details 🔗 GitHub: github.com/inclusionAI/LLaDA… 🔗 HuggingFace: huggingface.co/inclusionAI/L… 🔗 Tech Report: github.com/inclusionAI/LLaDA… #LLaDA #LLM #OpenSource #inclusionAI
1
9
2
37
2,752
Kata Containers 4.0 is now available! With this release, the Rust-based runtime becomes the project's default, delivering memory safety and the performance needed to sandbox AI agents at scale. Thank you to the community for making this release happen. #KataContainers
2
2
5
226
gnawux retweeted
🚀 Gemini 3.6 Flash and Gemini 3.5 Flash Lite are now live on ZenMux! Fast, capable, and ready for real-world workflows from lightweight tasks to high-speed multimodal applications. Try them now while the +20% extra credits offer is still live. Ends this month. 👇 zenmux.ai/google/gemini-3.6-… zenmux.ai/google/gemini-3.5-… #Gemini #GoogleAI #LLM #ZenMux
Today we’re expanding the Gemini family with three new models built to be faster, more token efficient, and reliable at scale. Meet the new Gemini models ↓
2
1
7
1,574
. @katacontainers' just released 4.0 early this morning (for better agent sandbox etc.). It has been 10 yr since I collaborated with @openinfradev & @sparkycollier. No one got more from the project and community than me. I'd like to say thank you to all the contributors. Move on!
1
1
4
977
Today we introduce LingBot-VA 2.0, the 🎆finale and also the capstone of LingBot launch week. 1️⃣ week, 6️⃣ models. Together they form our Embodied-Native Full-Stack — perception, world simulation, prediction, and action generation in one closed loop. One brain, many robots. 🔗 Project Page: technology.robbyant.com/ling… 🔗Paper: github.com/Robbyant/lingbot-… #inclusionAI #LingBot #EmbodiedAI
🤖Robots that think ahead and act in real time. LingBot-VA 2.0 — the first embodied-native foundation model. Not fine-tuned from a video generator. Built from scratch for the physical world. ✅ 93.6% success on bimanual tasks ⚡ 150 Hz single-GPU inference 🎯 20 demos to generalize. This is what happens when you stop adapting and start building natively. 🧵👇
1
7
785
Live at @icmlconf in Seoul! Join Ant Group for a cup of coffee, explore our latest research through our posters, and chat with Tech Leaders from inclusionAI. We’d love to talk about everything around the @AntLingAGI Ling Model series, LLaDA, AI safety, reasoning, and post-training. Come say hi! #ICML2026 #AGI #llm
2
1
11
756