Tongyi Lab

This account is verified because it's an official organization on X. Learn more

@Ali_TongyiLab

We advance the development of ASI and foster open source collaboration towards a smarter future.
Discord: discord.gg/mnPyh8ZUEc

careers-tongyi.alibaba.com/?lang=zh Joined March 2025

26 Following 26K Followers 1.2K Posts

Latest posts 5

Pinned post

Tongyi Lab @Ali_TongyiLab Jun 1

Introducing Qwen3.7-Plus, the latest flagship addition to the Qwen3.7 model series.

Built to bridge the gap between visual perception and terminal execution, it serves as a versatile foundation to power your diverse multimodal agent workflows. 🚀

Key highlights:

• Multimodal Interactive Hybrid Agent: Enables unified GUI & CLI operation across visual and text tasks.

• Versatile Coding Agent & Productivity Assistant: Handles full-modality input to supercharge your daily productivity.

• Visual Agent: Deepens agent intelligence with advanced perception, reasoning, grounding, and search-augmented QA.

• Cross-Harness Generalization: Delivers consistent, robust performance across diverse agent frameworks.

The next generation of Qwen3.7 family has arrived to support your AI agent workflows.

Tongyi Lab @Ali_TongyiLab Jul 24

Hello, creators and builders,

This week was packed with major drops:
Qwen3.8 (2.4T parameters) is coming open-weight, with Qwen3.8-Max-Preview now available on Alibaba’s Token Plan.

Qwen-Image-3.0 brings a real productivity tool with rich content, authentic details, and deep knowledge.

Qwen-Audio-3.0-TTS launches with production-grade speech synthesis across 16 languages. And Zvec 0.6.0 is live with faster retrieval and simpler deployment.

Let’s dive in.
open.substack.com/pub/tongyilab/…

Tongyi Lab @Ali_TongyiLab Jul 24

Zvec 0.6.0 is live.

New Group-By search adds per-group Top-K. Random rotation boosts INT4 recall by 50+ points with QPS unchanged. FTS now supports 34+ languages via the UAX #29 standard tokenizer.

Better recall, smarter retrieval, broader language support.

zvec-ai Verified account @ZvecAI Jul 24

🎉 Zvec 0.6.0 Release is Live!

This release brings major upgrades across retrieval, compute optimization, deployment experience, and the quantization framework. Upgrade and give it a try!

✨What's new:

• 🔍Retrieval: New Group-By search with per-group Top-K support, covering Flat, HNSW, and HNSW-RaBitQ indexes

• 📝Full-Text Search: Introduces a UAX #29 standard tokenizer, UTF-8 support, ASCII folding, and Snowball stemming for 34+ languages—a major upgrade for multilingual retrieval; Conjunction queries are now 22–38% faster with block-max pruning

• ⚡ Compute Optimization: INT8/INT4 quantization now supports random rotation, evening out the distribution to reduce variance and quantization error; INT4 recall improves by up to 50+ percentage points while QPS remains essentially unchanged

• 💾 Deployment Experience: DiskANN and libaio are now dynamically decoupled via dlopen, removing the libaio build dependency and plugin .so; libaio is automatically detected at runtime, with a graceful fallback when unavailable

• 🧩 Quantization Framework: The internal quantization module has been refactored into a pluggable Turbo framework, decoupling quantization logic from index code; future quantizers such as PQ and RaBitQ can be added independently without modifying index internals

📚 Learn more:

• 📄 Release Notes: zvec.org/en/blog/2026-0…
• 🧭 Roadmap: github.com/alibaba/zvec/i…
View quoted post on X

Tongyi Lab @Ali_TongyiLab Jul 23

The Qwen model family cyber atlas: a visual walkthrough of how AI really evolved — from large language models to embodied intelligence.

It all begins with large language models.

Watch the full video below.

3:38

Tongyi Lab @Ali_TongyiLab Jul 22

The next evolution of AI image generation is here! 🎨✨ Qwen-Image-3.0 is all about one word: "Real".

Moving beyond just pretty pictures, it now handles 4.5k token prompts, complex multi-grid layouts, 10px legible text, and even full LaTeX pages. A true productivity powerhouse for design, education, and e-commerce.

Go create. 🎨

Qwen Verified account @Alibaba_Qwen Jul 22

🎨 Meet Qwen-Image-3.0 — the third generation of our foundational image generation model.

If 1.0 was about "Precision," and 2.0 added "Variety, Completeness, Beauty & Authenticity," then 3.0 comes down to a single word: Real (实).

Three dimensions of "Real":
📰 Rich Content — prompts up to 4.5k tokens. One-pass generation of complex layouts: newspapers, storyboards, exam papers — even a 3×3 infographic grid or picture-in-picture-in-picture UIs.
🔬 Authentic Details — text legible down to 10px, full LaTeX paper pages, pores, hair strands & near-photographic skin texture.
🌏 Deep Knowledge — native rendering in 12 languages, 100+ art styles, realistic UIs (web / games / livestreams), plus world knowledge & live web retrieval.

Not just "good-looking" — genuinely useful. Image generation as a real productivity tool for design, content, education & e-commerce.

Go create 🏃🎨

💬Qwen Chat: chat.qwen.ai/?inputFeature=…
📝Blog: qwen.ai/blog?id=qwen-i…

View quoted post on X