TECH

Alibaba Releases Qwen 3.6-Plus AI Model With Enhanced Coding Capabilities

Published: Apr. 2, 2026 11:04 p.m. GMT+8

00:00

00:00/00:00

Listen to this article 1x

Alibaba’s Qwen released its next-generation large language model, Qwen 3.6-Plus on April 2. Photo: VCG

Alibaba Group Holding Ltd. on Thursday released Qwen 3.6-Plus, a new large language model it says markedly improves automated coding and AI-agent capabilities, as the company tries to regain momentum in the fast-moving artificial-intelligence race.

Alibaba is pitching the update around “agentic coding,” saying the model can break down complex programming tasks, write and test code, and troubleshoot iteratively until completion. As a native multimodal model, Qwen 3.6-Plus can also generate front-end web pages and produce code from screenshots, design drafts and text prompts.

You've accessed an article available only to subscribers

Subscribe today for just $.99.

VIEW OPTIONS

Unlock exclusive discounts with a Caixin group subscription — ideal for teams and organizations.

Subscribe to both Caixin Global and The Wall Street Journal — for the price of one.

Share this article

Open WeChat and scan the QR code

DIGEST HUB

Digest Hub

Back

Explore the story in 30 seconds

Alibaba released Qwen 3.6-Plus, enhancing agentic coding, multimodal code generation, and 1M token context.
Available on Bailian at 2 yuan (~$0.29)/M input, 12 yuan/M output tokens; rivals Claude Opus 4.5 on SWE-bench.
Follows AI reorganization via Token Hub amid competition from ByteDance, DeepSeek.

AI generated, for reference only

Who’s Who

Alibaba Group Holding Ltd.: Alibaba Group Holding Ltd. released Qwen 3.6-Plus, a multimodal LLM excelling in agentic coding, task breakdown, code generation from prompts/screenshots, and troubleshooting. It supports 1M-token context, rivals Claude Opus 4.5 on benchmarks, and is available on Alibaba Cloud’s Bailian (from $0.29/M input tokens). The firm restructured AI teams into Alibaba Token Hub to boost iteration amid competition from ByteDance and DeepSeek. (78 words)

ByteDance Ltd.: ByteDance Ltd. is a domestic rival to Alibaba in the AI race. In early 2025, it upgraded its closed-source Doubao 1.5 Pro model amid intensifying competition driven by DeepSeek’s R1.

DeepSeek: DeepSeek is a domestic AI rival to Alibaba. Its R1 model, released in early 2025, pushed competition toward test-time scaling amid intensifying rivalry.

Anthropic: Alibaba's Qwen 3.6-Plus matches Anthropic’s Claude Opus 4.5 in programming and agent benchmarks like SWE-bench and Terminal-Bench 2.0. It supports AI-agent frameworks including Claude Code for long-horizon tasks.

Minimax: Minimax, a startup, open-sourced its flagship models in early 2025 amid intensifying AI competition, as DeepSeek’s R1 advanced test-time scaling and ByteDance upgraded Doubao 1.5 Pro. This pressured Alibaba’s Qwen to differentiate in tool-calling and programming.

Moonshot AI: Moonshot AI, a startup, open-sourced its flagship models in early 2025, intensifying competition as Alibaba's Qwen series struggled with tool-calling and programming differentiation.

AI generated, for reference only