From the source

Hello Qwen2

New model

Alibaba announces the Qwen2 family of models, evolved from Qwen1.5, featuring five sizes, multilingual training, SOTA benchmarks, improved coding and math, and extended context length.

From the source
After months of efforts, we are pleased to announce the evolution from Qwen1.5 to Qwen2.
qwenlm.github.io

Other stories

Introducing QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. ✨
Elastic RL system for extreme-long-horizon agent training
Alibaba· 12 OctResearch paper
Qwen-Image-2.1: Compact, Efficient, and Unified Image Creation
Open-source image model unifies text-to-image generation and editing with 7B parameters.
Alibaba· 20 SepOpen-weight release
Qwen3.8-LiveTranslate: Names the speaker. Carries the meaning.

Announced: 06 Jun 2024 16:00 UTC
Detected: 03 Aug 2026 21:12 UTC (+788d after)

ModelsQwen2-0.5B, Qwen2-1.5B, Qwen2-7B, Qwen2-57B-A14B, Qwen2-72B

Visit source ↗qwenlm.github.io

Alibaba· 18 SepCapability change

Qwen3.8-Omni-Flash: Omni Senses. Agentic Delivery.

Audio-visual input pricing drops 93%, model improves 25% across omnimodal benchmarks.

Alibaba· 17 SepNew model

E-Commerce Bench: Long-Horizon Operations, Multi-Dimensional Evaluation

Alibaba· 3 Sep

Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving

Alibaba· 3 SepNew model

Hello Qwen2

From the source

Introducing QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. ✨

Qwen-Image-2.1: Compact, Efficient, and Unified Image Creation

Qwen3.8-LiveTranslate: Names the speaker. Carries the meaning.

Qwen3.8-Omni-Flash: Omni Senses. Agentic Delivery.

E-Commerce Bench: Long-Horizon Operations, Multi-Dimensional Evaluation

Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving