Ethical Stack.

Qwen-Image-2512

Apache 2.0 text-to-image checkpoint with English and Chinese metadata, local Diffusers pipeline, and links to Qwen Chat.

Score calculation

(3 + 1 + 7 + 6 + 7) ÷ 5 = 4.8

5 of six dimensions rated. Unrated dimensions are excluded from the average.

Full scoring method ↗
Setup details

Qwen-Image-2512 December 2025 checkpoint; hosted Qwen Chat and local weight inference are separate routes.

What fits your needs

Qwen-Image-2512 December 2025 checkpoint; hosted Qwen Chat and local weight inference are separate routes.

Text to image

The card describes the 2512 text-to-image update and a runnable Diffusers pipeline.

See the evidence

Text composition

Qwen reports improvements in text rendering and text-image composition for this release; results depend on prompts and settings.

See the evidence

Language route

The model card tags English and Chinese and links to Qwen Chat for hosted trials.

See the evidence
Tradeoffs to expect

Local compute

Weights are about 58GB; the official example selects CUDA/bfloat16 or CPU/float32 based on available hardware.

See the evidence

Hosted route terms

Qwen Chat uses Qwen Studio terms; these terms do not govern independent local deployments or every inference provider.

See the evidence
How to switch

Use weights or hosted chat

Run the official Diffusers pipeline locally, or open the Qwen Chat route linked from the model card and review its current account terms.

See the evidence
Practical questions

Training-data distinction

Qwen’s service-level summary describes company policies, not the exact training data/rights for this image checkpoint.

See the evidence

Reference images

The reviewed model release does not document a person-specific likeness-consent workflow; check rights to uploaded references and local law.

See the evidence

Qwen Chat route

For hosted prompts, check Qwen Studio account and data controls; the local checkpoint does not inherit them.

See the evidence

Energy practices

3/10↑

This grades disclosed operating practices, not the electricity or emissions of a particular generation.

See the evidence

The release gives limited runtime or control details but no comparable per-task footprint or electricity-source evidence. The official pipeline identifies CUDA or CPU execution routes and a large downloadable checkpoint. The demo switches dtype and device based on available hardware; batch size, image resolution and inference settings are user controls, but release documentation does not quantify energy reductions.

1 + 1 reporting + 0 electricity + 1 controls = 3/10. This grades documented practices and transparency.

Per-task energy, serving region, facility electricity mix, water and independent footprint review are not established in the cited release sources.

Reporting1/3
The official pipeline identifies CUDA or CPU execution routes and a large downloadable checkpoint.
Electricity sourcing0/3
The release materials do not report matched regional electricity sourcing or carbon-free supply for training or inference.
Resource controls1/3
The demo switches dtype and device based on available hardware; batch size, image resolution and inference settings are user controls, but release documentation does not quantify energy reductions.

Water disclosure

Water reporting has not been assessed for this choice.

Training data

1/10↑

The model release does not provide a complete inspectable training-source permission record.

See the evidence

The reviewed release documentation the checkpoint card links technical work but does not enumerate the training source inventory or permission basis. It provides no source-level permission ledger or creator remedy; a model-weight license governs downstream use, not the material used to train the model.

1 base + 0 source transparency + 0 documented permission + 0 creator controls = 1/10.

Named source inventory, source-level licenses, creator remedies, and independent verification remain unestablished for this release.

Source transparency0/3
The checkpoint card links technical work but does not enumerate the training source inventory or permission basis.
Documented permission0/4
The release sources do not publish a training-source permission basis or rights ledger; the weight license doesn’t settle training-data permission.
Creator controls0/2
No training-source creator permission-before-use, opt-out, removal or retraining route is documented for this release.
Data-rights rubric ↗

Your data & privacy

—Not enough evidence↑

Privacy has no single score for these downloadable weights because handling depends on the inference host and integrations.

See the evidence

The release can be run locally, but it also appears in hosted provider routes; those operators set their own request retention, training and region terms. A cloud API, where offered, is a separate policy route and is not inherited from the local weights.

A selected host, network path, telemetry, logging, external prompt enhancement, and data retention need route-specific review.

Openness

7/10↑

Apache 2.0 weights and pipeline instructions are public; the release lacks a source-level training data manifest.

See the evidence

The checkpoint is downloadable under Apache 2.0 and has a runnable Diffusers pipeline with a public model card. The full training data and training recipe are not disclosed in that card. This rating concerns release artifacts and reproducibility, not training-data permission.

7/10 reflects permissive weights and reproducible inference, limited by missing training data and full recipe.

The exact model checkpoint’s training-source inventory, source-specific licenses, and complete training pipeline remain unestablished.

Worker protections

6/10↑

Alibaba publishes group workforce indicators, an Employee Rights Code, employee reporting channels and compliance processes; these describe Alibaba employees, not the AI data-labeling supply chain.

See the evidence

Alibaba’s ESG employee-development pages report workforce composition and employee participation indicators, while its integrity and governance pages describe code compliance, reporting channels, compliance processes, and board oversight. These are concrete group-level employee systems, but public detail on annotation, evaluation, content moderation, and infrastructure contractor conditions or outcomes remains limited. These findings concern the developer or operator. This product’s use in a particular deployment needs separate evidence.

1 + 2 pay & conditions + 2 worker voice & remedy + 1 evidence & follow-through = 6/10. This grades documented worker safeguards, not a measured rate of mistreatment.

The reviewed pages don’t show AI-data subcontractor coverage, worker-level wage/hour outcomes, case counts or remediation outcomes for labor reports, or independent labor-audit results.

Pay & conditions2/3
Company ESG disclosures report workforce composition and employee participation indicators and describe employee-rights standards; this is concrete employee evidence but does not demonstrate coverage of AI data work or contractors.
Worker voice & remedy2/3
Alibaba describes employee reporting and escalation channels, privacy/non-retaliation protections, and dedicated compliance processes; public labor-specific case outcomes were not located.
Evidence & follow-through1/3
Governance materials describe compliance training and audit/control processes, but the reviewed public disclosures do not provide independent worker-level labor assurance or remediation results.
Workers rubric ↗

Military ties

7/10Not established↑

Qwen3.8’s released model licenses don’t spell out a military-use ban. Alibaba publishes stronger rules for its separate QwenWork hosted service; no Qwen3.8 military contract was established.

See the evidence

The Qwen3.8-27B release uses Apache 2.0, while the larger Qwen3.8 release has its own license. Neither reviewed release license carries the QwenWork hosted-service military restrictions. QwenWork prohibits specified weapons and military end uses and permits service enforcement, but that separate contract doesn’t spell out controls on downloaded Qwen3.8 weights. No Alibaba military contract or Qwen3.8 deployment was established in this bounded source review. These findings concern the developer or operator. This product’s use in a particular deployment needs separate evidence.

7 base + 0 safeguard adjustment = 7/10 (1 of 9 safeguard points; every 3 adds 1). No military relationship was established in the cited sources. This ranks above known defense ties without claiming a verified clean record.

The reviewed QwenWork contract doesn’t show an Alibaba defense relationship or a companywide model restriction. Self-hosted Qwen3.8-27B and Qwen3.8-Max use separate licenses; undisclosed contracts, deployments, and enforcement outcomes remain unknown.

Use restrictions0/4
The reviewed Qwen3.8 release licenses do not impose a specific military-purpose ban. Restrictions in the separate QwenWork customer contract are not applied to these weights.
Enforcement & oversight0/3
Hosted-service suspension rights don’t show monitoring or enforcement over downloaded Qwen3.8 copies. No release-specific military oversight was established.
Relationship disclosure1/2
Alibaba publishes the release licenses and a separate hosted military-use policy, making the scope distinction inspectable. No military customer or deployment record is established.

The QwenWork hosted-service terms include product-specific use safeguards, but no Alibaba defense contract or military deployment was established in reviewed materials. Its Qwen3.8 model repositories have distinct licenses, so the QwenWork safeguards are not attributed to self-hosted weights or treated as evidence of a military relationship.

Military rubric ↗

BDS targets

Not listed↑

We did not find Alibaba Cloud / Qwen team among the named targets in the two official BNC campaign lists. This is a limited list check, not BDS approval.

See the evidence

Alibaba Cloud / Qwen team is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Running downloaded weights locally is a different decision from paying the developer or a cloud host; check the host as well. These findings concern the developer or operator. This product’s use in a particular deployment needs separate evidence.

This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.

More from Alibaba

All sources

Side by side