Spaces:
Running
A newer version of the Gradio SDK is available: 6.24.0
title: Free Qwen 3.8 Max
emoji: 👀
colorFrom: purple
colorTo: blue
sdk: gradio
sdk_version: 6.19.0
python_version: '3.13'
app_file: app.py
pinned: false
short_description: A Space to try the new Qwen 3.8 Max for free
Free Qwen 3.8 Max
Try Qwen 3.8 Max with text and image inputs. Thinking mode is enabled by default; its reasoning summary appears in a collapsible panel while the final answer streams token by token.
The Space also includes a box-prompting playground inspired by Roboflow's Qwen 3.8 Max evaluation: click and drag positive and optional negative bounding boxes directly over an image, then ask Qwen to find similar objects. Three ready-made presets demonstrate shape, color, and size matching; each loads an editable instruction and example boxes. Reasoning and JSON tokens stream separately, then detection results are parsed from normalized XYXY coordinates and rendered back onto the original image.
A separate zero-shot object-detection tab accepts a plain-language detection prompt—or a Detect everything option—and renders class-labeled boxes without requiring examples. It includes ready-made shape, vehicle, and product-inventory prompts and uses the same streamed reasoning and JSON presentation.
This is an unofficial community demo, generously backed by a limited pool of Qwen API tokens. Availability is best-effort and the demo may be paused when the pool is exhausted.
Configuration
The Space expects one secret:
DASHSCOPE_API_KEY: a DashScope International API key with access toqwen3.8-max
Local development
uv sync
DASHSCOPE_API_KEY=... uv run python app.py
The app limits request concurrency, image size, queue depth, and per-session request frequency to keep the public token pool useful for more people.
Credits
Qwen and DashScope are trademarks or services of their respective owners. This Space is not an official Qwen or Alibaba product.