Alibaba launched Qwen3.7-Plus: multimodal agent with video analysis and autonomous coding

Alibaba has released an updated agentic model Qwen3.7-Plus on the Bailian platform. It understands images and videos, can invoke external tools, and iterates autonomously: writing code, running it, analyzing errors, and rewriting until achieving the desired result.

What is Qwen3.7-Plus

Qwen3.7-Plus is an intermediate release in the Qwen3 line, designed for agentic scenarios at the enterprise level. In 2026, Alibaba has already released several Qwen3 variants: compact models for local deployment, versions with extended context, and tools with chain-of-thought reasoning in Thinking mode.

The "Plus" suffix adds multimodality and a full agentic stack to the base architecture while maintaining the family's strong positions in reasoning and coding benchmarks. The model is available through Bailian — Alibaba Cloud's AI stack. The platform combines a model repository, RAG tools, an agentic application builder, monitoring, and deployment in a single interface.

In positioning, Bailian is the closest analog to Amazon Bedrock or Azure AI Studio, oriented toward enterprise customers in the Asia-Pacific region.

Four New Capabilities

The Qwen team highlights four key updates:

• Vision — the model sees images and video sequences. The agent works with interface screenshots, technical diagrams, and video demonstrations without external parsers or separate modal pipelines.

• Deep reasoning — extended step-by-step reasoning for tasks requiring chain-of-thought analysis. The model explains its reasoning process, which increases transparency of solutions and simplifies debugging of agentic systems.

• Tool invocation — native invocation of external tools, APIs, and databases directly from the request context. The model itself decides when and what to invoke, without manual orchestration wrappers.

• Autonomous iteration — an independent loop of "write → execute → analyze → fix". Qwen3.7-Plus doesn't just generate code: it runs it, reads error output, and rewrites it cyclically until achieving a correct result.

It is autonomous iteration that distinguishes Qwen3.7-Plus from a regular chat assistant: the model doesn't describe actions, but performs them in an environment, responding to real feedback.

Bailian and the Enterprise Bet

Launching on Bailian is a deliberate strategic choice. The platform is already integrated with ERP, CRM, and industry-specific databases of Alibaba Cloud's major clients. An agentic model with native tool invocation connects to these sources without additional orchestration, lowering the adoption barrier and reducing time to first result. Alibaba positions Bailian as infrastructure for enterprise AI automation. Qwen3.7-Plus becomes the agentic core of the platform: it sees documents and data, reasons, invokes necessary services, and achieves results through iterations. The bet on its own ecosystem is a key advantage over Western competitors, who find it harder to integrate with the corporate systems of Asian clients due to regulatory barriers and local market specifics.

What This Means

The release of Qwen3.7-Plus is part of the general 2026 trend: AI labs are transitioning from conversational models to agents capable of autonomously executing long tasks in real systems. Alibaba competes with Anthropic, OpenAI, and Google by betting not on the model itself, but on its integration into enterprise infrastructure. For business, this means that complex multi-step tasks — analyzing documents with images, working with corporate APIs, autonomous coding — become available without hiring a specialized AI team.