Qwen Image 2.1 Pro Review: Uncensored Pro-Tier Images and Edits

The gallery holds published examples from the Qwen Image 2.1 Pro model page, all made with prompt expansion off at the 2k tier. Everything in it shows clothed adults, suggestive at most; our content-test images are not shown, and we report those results in words.

One light source in a storm. A woman stands in the brass-and-glass lamp room of a lighthouse at night in an oversized cream sweater slipping off one shoulder, rain on the glass and the beam sweeping behind her. A single oil lamp on the left lights her face, as the prompt asked.

Two people, one moment. In a mirrored tango studio, a couple in their thirties dance close, her black dress slit to the thigh, his waistcoat over a white shirt with the sleeves rolled, the mirrors repeating them into the dark. One detail drifted: the prompt wrapped her leg around his, and the image keeps both her feet on the floor.

Lettering. A magazine cover on an oxblood backdrop: "AFTER DARK" in large ivory serif capitals and "THE SILK ISSUE · NO. 07" in small letters, both spelled exactly. The cover line sits just under the masthead rather than at the lower left where the prompt put it.

A man at dawn. In an empty boxing-gym locker room, a shirtless man with a towel round his neck wraps his knuckles on a worn bench, grey light from a high window on the right.

Edits from one source. From a photo of a singer in a velvet dress, the edit changed the dress to an ivory silk robe and moved her to a hotel bathroom with a glass of champagne, warm sconces and a mirror behind her, keeping her face and hair.

Edits from two sources. A man from one photo and the singer were combined at a candlelit table on a vineyard terrace at dusk, him in his open white shirt, her in a backless emerald dress. Both faces stayed recognisable.

Our quality set. On 14 October we also ran the 14 questions we use for every image model's blind test (8 text-to-image, 6 edits), at the 2k default with prompt expansion off. Six diners at a garden table were six, both text questions were spelled correctly, and the six edits kept what they were told to keep. The misses were the ones listed above: four fingers for three, a half-full glass, a balloon in the wrong colour and a watercolor edit that stayed close to the photo.

Sizes. The 2k tier gave 2048 x 2048 at 1:1, 2688 x 1536 at 16:9 and 1760 x 2368 at 3:4; 1k at 2:3 gave 832 x 1248. Edits without aspect_ratio kept the source's shape: a 2:3 source came back at 1664 x 2496 and a wide source at 2704 x 1536.

Write a short brief, not a keyword list: the medium first ("Cinematic photograph of…"), then the person with an age, the wardrobe and the setting, then one light source and a lens or film note. Put any text you need in double quotes and say where it goes, and check the position in the result. For edits, name each source by its number ("Image 1"), say what changes, and say what must stay. Keep prompt expansion off so your words reach the model as written, and keep every person an adult.

1. Text-to-image, one lamp in a storm. aspect_ratio 3:4, resolution 2k.

Cinematic photograph of an adult woman in her early thirties standing in the brass-and-glass lamp room of an old lighthouse during a night storm, an oversized cream knit sweater slipping off one bare shoulder, bare legs, damp auburn hair, one hand on the cold glass. The rotating beam sweeps behind her through sheets of rain, her face lit warm by a single oil lamp on the left, the sea black and white with foam far below. 35mm lens, shallow depth of field, fine film grain.

Run this prompt on Qwen Image 2.1 Pro →

2. Text-to-image, a magazine cover with exact lettering. aspect_ratio 2:3, resolution 2k.

Fashion magazine cover photograph on a deep oxblood backdrop: an adult woman in her thirties in an open black satin tuxedo blazer worn over a black lace bralette, hands in her trouser pockets, chin lifted, cool direct gaze, soft beauty dish from above. At the top in large ivory serif capitals the masthead reads "AFTER DARK". Lower left in small ivory sans-serif letters: "THE SILK ISSUE · NO. 07". No other text.

Run this prompt on Qwen Image 2.1 Pro →

Both lines came back spelled exactly; the second one moved up under the masthead. If the position matters, say it twice or run several images.

3. Edit, new wardrobe and setting from one photo. One source in image_urls, aspect_ratio left unset.

Keep the woman's face, hair and expression from Image 1. Change her velvet dress into a loosely tied ivory silk robe slipping off one shoulder and place her in a marble hotel bathroom at night, leaning against the vanity with a glass of champagne, warm sconce light, a fogged mirror behind her, photorealistic.

Run this prompt on Qwen Image 2.1 Pro → Add your own portrait as the source image.

4. Edit, two people from two photos. Two sources in this order, the man then the woman; aspect_ratio 3:2, resolution 2k.

Put the man from Image 1 and the woman from Image 2 at a candlelit dinner table on a vineyard terrace at dusk. He keeps his open white shirt; she wears a backless emerald silk dress instead of her velvet dress. She leans across the table toward him, his hand on her wrist, both smiling, rows of vines and a violet sky behind them. Keep both faces recognisable, photorealistic, warm candle light.

Run this prompt on Qwen Image 2.1 Pro →

Weak: "three fingers and two" and trusting the count, or "turn it into a watercolor" and expecting a painting. Strong: describe each hand or object you need on its own, and for heavy restyling describe the brushwork, the paper and the washes you want to see.

SpicyAPI adds no content filter, and the page label is Uncensored. On 14 October 2026 we tested both endpoints in two rounds, with adult subjects only, and checked every image by eye. The levels are L1 Suggestive (lingerie, a slipping strap), L2 Partial, L3 Strong, L4 Intense (an intimate scene between two adults) and L5 Extreme, in two kinds: consensual restraint between adults, and cinematic gore. Readings below are delivered / softened / refused.

Our standard test, at the defaults: 36 of 36. Six probe types with three runs each per endpoint, prompt expansion off, thinking on, 2k text-to-image and edits from our own source photos:

For restraint we counted a run only when both the restraint and the strike, or its marks, were drawn; for gore, only when the wound itself was drawn. Every run met that bar. Two of the restraint images drew the woman as a man, which is a quality slip, not a softened result.

The first round agrees. Earlier the same day, seven adult requests with expansion off all came back as asked: L3 content for a woman and for a man, an intimate scene drawn in full, a restraint scene, a battlefield wound, and two edits that took a top off and went to L3 content from a clothed photo.

Prompt expansion: one run. We sent one Pro request with enable_prompt_expansion on, at L3, and it still came back as asked. On Turbo, the same rewriter softened several adult requests and returned two edits unchanged, so one run is not enough to say Pro behaves differently. Keep it off for adult work, which is the default here.

Across all 65 requests we sent on 14 October, including 14 safe-for-work quality questions and the gallery examples, nothing was refused. This section's gallery shows two published examples at the suggestive end; the content-test images are described here and not shown.

Not tested yet: an illustration or anime style instead of photography, scenes with two men or two women, repeat runs with expansion on, and transparent output.

Adults only, in every prompt and every source photo. Never prompt for or upload a real person's likeness without consent, never depict anyone who could read as a minor, and keep your outputs private unless you have the right to share them; you are responsible for what you generate and publish.

• The same request for less? Qwen Image 2.1 Turbo takes exactly the same parameters at $0.0144 an image, with open weights and 8 steps. In our blind test it scored 71.4 to Pro's 72.4: level on edits, a little behind on text-to-image image quality. See our Qwen Image 2.1 Turbo review.
• A mask or a format choice? Qwen Image 2.1 is the open-weight base model; on SpicyAPI it adds a mask for local edits, a choice of output format and a background setting, priced by size and returning one image per request. See our Qwen Image 2.1 review.
• Signs and long text in the picture? Qwen Image 3.0 Pro is built for lettering, priced by size. See our Qwen Image 3.0 Pro review.
• Your own LoRA on the 2.1 base? Qwen Image 2.1 LoRA takes your own weights files with a strength for each. See our Qwen Image 2.1 LoRA review.