---
title: "Qwen Image 2.0 Blind Test Results: Scores by Use Case"
canonical_url: "https://workroom.everypixel.com/models/qwen-image-2-0"
score: 64.5
rank: 3
scale: "Anchor comparison of generative image models, v3"
tested: 2026-08-25
revised: 2026-09-14
---

# Qwen Image 2.0 Blind Test Results: Scores by Use Case

**Score:** 64.5. A real photograph is 100.
**Rank:** #3 on the scale
**Author:** Dmitry Shironosov (Head of AI Research, Everypixel Workroom)
**Tested:** 2026-08-25 · revised 2026-09-14
**Canonical URL:** https://workroom.everypixel.com/models/qwen-image-2-0

## Independence

Workroom is an AI router: we give access to third-party models, including Qwen Image 2.0, and we are not affiliated with Alibaba. We have no stake in any single vendor winning; the router is only useful if it sends each brief to the right model. The judgement-level data behind this card is published by Everypixel research under CC BY 4.0 and linked below, so every number here can be checked independently.

## TL;DR

Qwen Image 2.0 scores 64.5 on our photo-anchored scale, where a real photograph is 100; it is the middle rung of the ladder. In blind pairwise tests on 48 commercial scenes it clearly beat FLUX.2 Klein, fell clearly behind GPT Image 2 and the reference photographs, and showed no measurable difference from Seedream 4.0, leaning ahead on product shots and typography.

## Where it stands

| Rank | Model | Score | 95% interval |
|---|---|---|---|
| #1 | GPT Image 2 | 104 | — |
| — | Real photograph | 100 | — |
| #2 | Qwen Image 2.0 Pro | 80 | 68.6–93.8 |
| #3 | Qwen Image 2.0 (this card) | 65 | — |
| #4 | Seedream 4.0 | 57 | 44–70 |
| #5 | FLUX.2 Klein | 24 | — |

Axis: believability. A real photograph is 100. A score above 100 is the scale's technical ceiling and does not mean parity.

### By use case

| Category | Score | Rank |
|---|---|---|
| Interior | 104 | #3 |
| Product and e-commerce | 87 | #3 |
| Street and architecture | 73 | #3 |
| Social marketing | 71 | #4 |
| Fashion and beauty | 69 | #3 |
| Food and beverage | 46 | #4 |
| People and lifestyle | 15 | #4 |
| Layout and typography | 14 | #3 |

*Category positions come from 6 scenes each — a tendency, not a measurement. A real photograph is 100 on every ruler.*

## When to use as default

Use Qwen Image 2.0 as the everyday mid-tier model for product shots, layouts with typography and street or architecture scenes: on those categories it leaned ahead of Seedream 4.0, and overall it sits well above FLUX.2 Klein on believability.

## When to use as finisher

Not the finisher for the most believable result: GPT Image 2 was preferred over it in 72.0% of judgements (vote share, ties split), with a believability gap of 0.83. Draft in Qwen Image 2.0 and finish the hero asset in GPT Image 2.

## When to look elsewhere

Do not expect photographic parity: the reference photographs were preferred over Qwen Image 2.0 with a clear believability gap of 0.74. For social-marketing and fashion briefs Seedream 4.0 leaned ahead in our test, so a side-by-side take is worth it.

## Direct blind tests

| Opponent | Result | Gap (Bradley–Terry, 95% CI) | Preferred in | Judgements |
|---|---|---|---|---|
| FLUX.2 Klein | Qwen Image 2.0 was preferred over FLUX.2 Klein in 69% of blind comparisons. | +0.85 (interval not published) | 68.6% (ties split, all questions, vote share) | 144 |
| GPT Image 2 | GPT Image 2 was preferred over Qwen Image 2.0 in 72% of blind comparisons. | −0.83 (interval not published) | 28.0% (ties split, all questions, vote share) | 144 |
| Real photograph | The real photograph was preferred over Qwen Image 2.0 in 62% of blind comparisons. | −0.74 (interval not published) | 37.6% (ties split, all questions, vote share) | 143 |
| Qwen Image 2.0 Pro | Qwen Image 2.0 Pro was preferred over Qwen Image 2.0 in 47% of blind comparisons. | −0.45 [−0.72, −0.21] | 24.8% (wins over all judgements, believability only, vote share) | 432 |
| Seedream 4.0 | A coin flip: 52% for Qwen Image 2.0, no measurable difference. | +0.07 [−0.14, +0.28] | 51.8% (ties split, all questions) | 432 |

## Price

| Provider | Price | Checked |
|---|---|---|
| [Alibaba Cloud Model Studio](https://www.alibabacloud.com/help/en/model-studio/model-pricing) | $0.035 / img | 2026-09-25 |
| [fal.ai](https://fal.ai/qwen-image-2.0) | $0.035 / img | 2026-09-25 |
| [Replicate](https://replicate.com/qwen/qwen-image-2) | $0.035 / img | 2026-09-25 |
| [Runware](https://runware.ai/docs/models/alibaba-qwen-image-2-0) | $0.035 / img | 2026-09-25 |

Average: $0.035 / img across 4 providers

Prices are for one 1024×1024 image at standard settings, checked 25 September 2026; the price at the top of the page is the average of the providers listed. Workroom pricing is on the pricing page.

## License

Qwen Image 2.0 is Alibaba's model, available through the Alibaba Cloud Model Studio API. Earlier Qwen-Image releases were published with open weights under Apache 2.0; check Alibaba's current terms for version 2.0.

## FAQ

### Is Qwen Image 2.0 better than Seedream 4.0?

Not measurably. On believability the gap is +0.07 with a 95% interval from −0.14 to +0.28, which includes zero, and Qwen Image 2.0 was preferred in 51.8% of judgements with ties split. The difference is in the mix: Qwen leaned ahead on product shots, typography and street scenes, Seedream on social and fashion.

### Why is Qwen Image 2.0's score shown without a range?

Qwen Image 2.0 is a rung of the ladder: its score was fitted when the scale was built and is frozen for later sessions. The published baseline gives the fitted scores of the rungs without score intervals, so the card shows the point value only rather than an invented range.

### How big is Qwen Image 2.0's gap to GPT Image 2?

Clear. The believability strength gap is 0.83 in GPT Image 2's favour, and the dataset notes that the lower rungs separate reliably. The same gap was re-measured in the September session and the frozen value fell inside that session's interval. Across the three questions GPT Image 2 was preferred in 72.0% of judgements.

### How was Qwen Image 2.0 tested?

It was one of four rungs compared blind on 48 commercial scenes in August 2026, paired on every scene with FLUX.2 Klein, GPT Image 2 and the reference photographs, each pair judged three times on three questions. In September it was also paired with Seedream 4.0 on the same scenes.

### Is Workroom affiliated with Alibaba?

No. Workroom is an AI router that gives access to third-party models, including Qwen Image 2.0, and is not affiliated with Alibaba. The judgement-level data behind this card is published by Everypixel research under CC BY 4.0, so every number on this page can be checked against the source dataset.

## Methodology

Between 21 and 25 August 2026 we compared four rungs of our quality ladder blind on 48 commercial scenes: FLUX.2 Klein, Qwen Image 2, GPT Image 2 and the real photographs each scene was written from. Every rung was paired with every other on every scene — 288 pairs and 863 judgements — and each pair was judged three times, without any indication of which image came from which model, on three questions: which image follows the prompt better, which is more believable as a real photograph, and which is more attractive. In September Qwen Image 2 was also paired blind with Seedream 4.0 on the same scenes: 432 more judgements.

- Scenes: 48; generations per scene: 1
- Compared with: FLUX.2 Klein, GPT Image 2, Real photograph
- Pairs: 288; judgements: 863; each pair judged 3 times
- Collected: 2026-08-21 … 2026-08-25
- Questions: “Which image follows the prompt better?” “Which is more believable as a real photograph?” “Which is more attractive?”
- Scale: 0.020835 strength per point; A real photograph is 100.
- Category positions are fitted from the same judgements, 6 scenes per category, on the same scale.

### Caveats

- Prompt adherence is biased against the photograph: each prompt was written by looking at the photograph, and a model executes text literally, while a photograph always contains incidental detail the description does not mention. Raters count that as a miss — which is why the absolute scale is built on believability alone.
- The share of ties is inflated: during collection the “equal” option was pre-selected, so a criterion the rater never touched was recorded as a genuine tie. The defect was fixed on 28 August 2026, after this data was collected. Tier ordering survives it: dropping every tie only widens the gaps.
- All images were shown at 2048 px on the long edge — native output size for the generative tiers; the photographs were downscaled from 30–45 megapixels. On screen the photographs are not softer than the generated images.

Rater agreement is low by the usual yardstick (Fleiss' kappa 0.16–0.20). Half the pairs compare adjacent tiers, where “equal” is an honest answer. Conclusions hold at the level of the whole corpus, not for a single scene, pair or rater.

## Links

- Full study: https://research.everypixel.com/anchor-ladder/
- Votes (CSV, 106 KB, CC BY 4.0): https://workroom.everypixel.com/datasets/anchor-comparison-v3/votes.csv
- Results (JSON, 5 KB, CC BY 4.0): https://workroom.everypixel.com/datasets/anchor-comparison-v3/results.json
- Scenes and prompts (CSV, 30 KB, CC BY 4.0): https://workroom.everypixel.com/datasets/anchor-comparison-v3/scenes.csv
- README (MD, 2 KB, CC BY 4.0): https://workroom.everypixel.com/datasets/anchor-comparison-v3/README.md
- Images (ZIP, 128 MB, All rights reserved — verification only): https://wr.everypixel.com/ghost-research/geo-benchmarks/datasets/anchor-comparison-v3-images.zip

## Citation

Everypixel research (2026). Anchor comparison of generative image models, v3 [Data set]. Ratings collected 21–25 August 2026.

Data license: CC BY 4.0. Images: All rights reserved; ten Unsplash photographs under the Unsplash License.

---
HTML version: https://workroom.everypixel.com/models/qwen-image-2-0
Last modified: 2026-09-14