Qwen-Image 2.0 Pro is Alibaba's pick: which Qwen ids does Sume list?

Alibaba's Model Studio page names qwen-image-2.0-pro as the recommended Qwen-Image model, but the Sume image model slugs in the repository are qwen/qwen-image and qwen/qwen-image-max. If you need 2.0 Pro specifically, confirm it against GET /v1/images/models before building on it, because a model id that is not in the catalog cannot be called.

Facts about Alibaba's lineup come from its Qwen-Image text-to-image API page, read 2026-10-02. Sume's list comes from the Image API and the image model slug list in the repository's API contract package.

What does Alibaba list on the page?

The page lists four text-to-image models: qwen-image-2.0-pro (marked recommended, with stronger text rendering, realistic texture and semantic adherence), qwen-image-2.0 (a balance of quality and speed), qwen-image-max (higher realism and fewer artifacts) and qwen-image-plus (diverse artistic styles). It says only qwen-image-plus and qwen-image support asynchronous calls.

How do the size and count limits compare?

Alibaba says the 2.0 series takes total pixels between 512x512 and 2048x2048, with 2048x2048 as the default, while the Max and Plus series default to 1664x928 and support ratios from 16:9 to 9:16. The Max series returns a fixed single image, and other series allow one to six.

Sume does not pass those vendor limits through. It publishes its own descriptors per model: aspect_ratio values, an n range and input_references range. Because the Sume page says every catalog model is served through one sume endpoint, the model-level and endpoint-level descriptors are identical, and the descriptor you read is the contract.

How do you confirm what is live today?

The repository tells you what the code knows about; the live catalog tells you what your key can call. Query it and filter for Qwen.

curl -s "https://api.sume.com/v1/images/models" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  | python3 -c "import sys,json;print([m['id'] for m in json.load(sys.stdin)['data'] if 'qwen' in m['id']])"

What should you do if you need 2.0 Pro today?

Two honest options. Use qwen/qwen-image-max on Sume and compare it on your own prompts, since Alibaba describes it as higher realism, or call Alibaba directly for the 2.0 series and keep the rest of your pipeline on Sume. Do not assume a vendor's recommended model id is available on a router; model names change faster than catalogs.

If you want an alert when a new id appears, diff the catalog on a schedule. For the question of Qwen-Image 3.0 see that post, and for the Max model's text-only limit see Qwen Image Max on Sume.

A short decision rule

Model catalogs on routers lag vendor launches by design, because each addition needs pricing, descriptors and testing. Treat the repo list and the live list as different things, and build your client so the model id is a configuration value, not a constant buried in code.

• If your use case is stylised art, Alibaba points at qwen-image-plus, which Sume's repo does not list, so you would call Alibaba for it.
• If you want realism and fewer artifacts, test qwen/qwen-image-max against your own prompts on Sume.
• If you want the recommended 2.0 Pro, first check the live catalog output from the command above.