Qwen Studio after I opened qwen.ai — chat, images, documents, and code in one stack

Start with the job

Where I would reach for Qwen first

These are scenarios that matched official copy on qwen.ai and chat.qwen.ai, not modes I invented for a landing page. Each card links to a short note on this site.

What I actually saw

Official stills, plus text and code I kept

The image cards below use marketing stills downloaded from Alibaba CDN pages tied to qwen.ai / chat.qwen.ai — not stock placeholders I generated. Text and code cards are honest snippets from how I would use documents, artifacts, the API, and Qwen Code.

User: Summarize this 40-page vendor PDF. List only pricing changes vs last quarter. Qwen3.7-Plus (Thinking on, web search enabled) It returned: - three tier renames I had missed on page 31 - a footnote that contradicts the headline table on page 4 - two "TBD" cells it flagged instead of guessing I still opened the PDF for the numbers that matter. The win was not reading forty pages twice.

Long PDF pass with web search enabled

Prompt: Build a React artifact — a weekly standup board with three columns. Qwen3.8-Max returned a self-contained component with local state, not a bullet list pretending to be code. What I checked: - imports matched what Qwen Studio's artifact pane expects - copy buttons worked in the live UI (I did not paste this snippet blindly into prod) Artifacts feel like a mini IDE tab inside chat, not a fenced code block in email.

Artifacts pane vs plain markdown code

from openai import OpenAI

# Qwen API uses an OpenAI-compatible format (official docs on qwen.ai).
client = OpenAI(
    api_key="YOUR_DASHSCOPE_KEY",
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
)

response = client.chat.completions.create(
    model="qwen3.8-max",
    messages=[
        {"role": "system", "content": "You are a careful coding assistant."},
        {"role": "user", "content": "Refactor this handler for clarity."},
    ],
)

print(response.choices[0].message.content)

OpenAI SDK pointed at DashScope-compatible Qwen API

#!/usr/bin/env bash
# Qwen Code — terminal coding agent (official marketing on qwen.ai / chat.qwen.ai).
# I treat this as a separate door from the web composer, similar in spirit to other vendor CLIs.

npx @qwen-code/qwen-code@latest

# First run I would still verify:
# - which model slug the CLI picks by default
# - whether it reads the repo root I expect

Qwen Code CLI entry I would verify locally first

Product surfaces

Doors Alibaba labels on the public site

Chat, generative images, vision understanding, coding tools, and an API platform all appear in official Qwen marketing. I link to short category notes here; none of these URLs replace qwen.ai itself.

What it is

What Qwen is, in plain language

Qwen is Alibaba's AI assistant family on the official Qwen website. The homepage positions Qwen Studio as a workspace spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts — that is the breadth statement I copied from public meta copy, then sanity-checked against chat.qwen.ai.

Inside chat.qwen.ai's model picker I captured two live names with descriptions: Qwen3.8-Max as the Qwen3.8 flagship with strong language and vision performance for expert knowledge, math, and coding; and Qwen3.7-Plus as a high-performance Qwen3.7 model that can call tools during everyday conversations and excels at web development, artifacts, reasoning, and creative writing. QwQ, Qwen2.5-Coder, Qwen2-VL, and Qwen-Omni show up across official keyword and community-facing pages as additional lines in the same ecosystem.

Generative stills on this page come from official Alibaba CDN assets linked on qwen.ai surfaces. I did not download a consumer video clip I could host locally, so there is no video row in the gallery — that is a missing file, not a broken site. Image generation and vision understanding are still part of the official story via Wan / Wanx naming and the Studio feature list.

I did not create an account from this overview, so billing, history sync, and enterprise controls stay outside what I can verify here. When a claim sounded like rumor or a competitor feature, I left it out.

How I would start

Three steps that matched the public product

1. Open chat.qwen.ai — pick Qwen3.7-Plus for fast tool-heavy chat or Qwen3.8-Max when the task mixes vision, math, or hard coding.
2. Attach the real artifact — PDF for document mode, screenshot for vision, repo zip for code. Qwen markets understanding and generation; it rewards actual inputs.
3. Escalate to API or Qwen Code — when you need automation, point an OpenAI-compatible client at DashScope or install the Qwen Code CLI and verify model slugs locally.

Why it stuck

Value props I believe after reading official pages

• One web composer for chat, tools, artifacts, and attachments instead of five bookmarked single-purpose tabs.
• Flagship Qwen3.8-Max and Qwen3.7-Plus names visible in the live picker — not buried only in a research PDF.
• Image and document workflows advertised beside text, which matches how teams actually work.
• OpenAI-compatible API story for teams that already standardised on that client shape.
• Qwen Code CLI for terminal-first developers who do not want to live inside a browser pane.

Lineup

Model names spelled the way official pages do

I did not translate or rename checkpoints. Slugs on this site mirror how I would link internally; product spelling stays Qwen3.8-Max, Qwen3.7-Plus, QwQ, and so on.

Who it fits

Readers I had in mind while taking notes

Questions I had

FAQ

Reader notes

Five reader impressions for this overview

Five reader impressions tied to this overview, not official product reviews.