vs Qwen

Qwen Studio has a Memory feature. What does it actually retain?

Alibaba Cloud's qwen.ai privacy policy describes a Memory feature that can retain details from your conversations for future chats. That is real continuity, but the published scope, controls, and regional handling deserve a closer look.

Does Qwen Studio remember previous conversations?

Alibaba Cloud (Singapore) Private Limited publishes the terms of service and privacy policy that govern Qwen Studio and the other services reached through qwen.ai, both dated May 19, 2026. Among the purposes it lists for using your data, the privacy policy includes providing personalized services "including through the Memory feature, which may retain certain details from your conversations to enable more personalised and consistent responses in future chats."

Qwen's own product page describes Qwen Studio as "an AI assistant for everyone, powered by the Qwen series models" and says it is "free to use, open to all." The privacy policy says account information such as your name and email address is collected when you sign up, and that you may instead log in with an existing third-party account such as Google or GitHub. Memory, as described, rides on that account.

Model and mode still matter, and so does naming. Alibaba's terms call the consumer surface Qwen Studio, while the Qwen team's Qwen3-Max-Thinking announcement describes Qwen Chat conversations in which that model can select built-in Search, Memory, and Code Interpreter capabilities. That is one documented case, not a published guarantee that every model, mode, plan, or region behaves the same way.

What happens to your Qwen conversations and uploads?

The privacy policy says Alibaba may collect the prompts and content you upload, including text, files, images, audio, and video, and that it may collect voice or video recordings when you use the voice, speech, audio, and audiovisual features. It notes that this can involve processing data that may count as biometric identifiers under applicable laws. It also says that rating an output, for example with the thumbs up or thumbs down icon, stores the related conversation as part of that feedback.

The terms of service go further. Alibaba says it may process any User Content to improve its services and develop new products and services, and that you "expressly authorise and consent" to it using and storing User Content that is not personal data to develop and improve its machine-learning and artificial-intelligence technologies, and to processing and storing that content outside the jurisdiction where you are using the services.

Alibaba's published training data summary states the position plainly: "There are instances when we use data provided by users, including user interactions with our services, to improve our services, in accordance with our relevant terms of service and/or privacy policy. Our users have the option to opt-out of their content being used for training by submitting their requests to us." Opting out is therefore described as a request you send, not a switch those pages document inside the product.

How do Qwen's international and mainland versions differ?

The qwen.ai terms say they govern "Qwen Studio and other services accessible via qwen.ai," made available by Alibaba Cloud (Singapore) Private Limited, and Alibaba processes copyright notices under the Copyright Act 2021 of Singapore. That is the international surface, and those are the documents that apply to it.

Alibaba's mainland consumer surface is a separate product on a separate domain: tongyi.com now redirects to qianwen.com, which presents itself there as Alibaba's all-round AI assistant. The qwen.ai terms and privacy policy do not cover that product, so its account rules, memory behavior, and data handling have to be read on its own pages rather than inferred from qwen.ai.

Region matters inside qwen.ai as well. The privacy policy says personal data is stored or processed in Singapore and Mainland China, and that certain entities in Alibaba's corporate group located in both places are given limited remote access under strict security controls and authorization approval protocols. For transfers out of the European Union, European Economic Area, Switzerland, or the United Kingdom, it says it relies on adequacy decisions or standard contractual clauses.

Can you export or delete what Qwen keeps?

The privacy policy lists the rights that may apply depending on where you live: access, erasure, correction, data portability, restricting or objecting to processing, withdrawing consent, and complaining to a data protection authority. It says you can exercise some of these rights through your services account, and that if you cannot, you should submit a request. Portability appears as a right to claim rather than a self-serve conversation export those pages describe.

Deletion has a stated tail. Alibaba says it retains personal data where it has an ongoing legitimate need, and that "in certain circumstances, we may need to retain your personal data after your account has been deleted," with periods depending on the data and the legal requirement. When there is no further legitimate need or legal reason, it says it will securely delete the data or anonymize it so it can no longer be associated with you.

Age is the sharpest boundary, and Alibaba's documents word it differently. The terms of service say you must be at least 18 years old or the age of majority in your jurisdiction, whichever is higher, and must not use the services if you are younger. The privacy policy says the services are not directed to or intended for individuals under 18. The usage policy phrases the same section as "Users must be 18+ or have parental consent." Where a vendor's own pages differ, the terms you accepted at sign-up are the ones that bind you.

How does Fostera approach continuity, and Qwen engines?

Fostera centers continuity on a named Soul rather than on a single thread. In practice that means selective Soul and project memory that you can inspect, edit, and delete, with relevance-based retrieval that can miss details. The memory browser is the point of the design: retained entries are listed individually, and Custom Rules give durable voice, canon, or working preferences an explicit home instead of leaving them to be inferred from an old conversation.

Fostera routes automatically among eligible engines, and manual selection is a Premium feature, so a Soul's retained context does not have to be rebuilt when the engine behind a reply changes. Projects group related chats, files, instructions, and project-scoped memories under that same Soul, and Fostera provides a structured JSON export of your account data from settings if you want a copy before you change anything.

Fostera routes some requests to eligible Qwen engines through Alibaba Cloud Model Studio, which Fostera's privacy page lists as a configured model and inference platform alongside its other providers. Alibaba's own retention and training terms apply to those requests, and Fostera does not override them. Connected provider terms vary, and memory deletion cleanup may take up to 30 days. Choose Qwen Studio when you want Alibaba's free consumer surface and its model family directly; choose Fostera when inspectable memory, project-scoped continuity, and continuity that survives an engine change matter more.

Qwen memory vs Fostera Souls

FeatureFosteraQwen
Cross-chat continuityselective Soul and project memory that you can inspect, edit, and delete, with relevance-based retrieval that can miss detailsMemory feature that may retain certain conversation details
Memory visibilityMemory browser with individually editable entriesMemory purpose stated in the privacy policy; no memory browser documented there
Data and training useConnected provider terms vary, and memory deletion cleanup may take up to 30 daysUser Content may be processed to improve services; training opt-out by request
Model portabilityFostera routes automatically among eligible engines, and manual selection is a Premium featureQwen series models inside Alibaba's own product
Age policyFostera is SFW for ages 13+, with confirmed guardian consent required for ages 13–17Terms require at least 18 or the local age of majority, whichever is higher
Free access3 Souls with flexible monthly capacity for conversations, media, creations, analysis, voice, and automationQwen Studio described as free to use and open to all

Frequently asked questions

Does Qwen have memory?

Alibaba's qwen.ai privacy policy describes a Memory feature that may retain certain details from your conversations to enable more personalized and consistent responses in future chats. It states that purpose rather than publishing a per-model retention contract, so behavior can differ by model, mode, plan, or region.

Why does Qwen still forget things?

The published wording is that Memory may retain certain details, not that every message is replayed. Long threads can also be truncated or summarized by the application, retrieval can miss a fact, and a different Qwen surface such as the API or Qwen Code carries context under its own rules.

Does Alibaba train on Qwen conversations?

Alibaba's training data summary says there are instances when it uses data provided by users, including user interactions with its services, to improve those services, and that users can opt out of their content being used for training by submitting a request. The qwen.ai terms separately authorize processing User Content to improve and develop its services.

Can I export my Qwen chat history?

Alibaba's privacy policy lists data portability among the rights that may apply depending on your jurisdiction, and says some rights can be exercised through your account while others require a submitted request. It does not describe a self-serve conversation export on that page.

How old do you have to be to use Qwen?

The qwen.ai terms of service say you must be at least 18 years old or the age of majority in your jurisdiction, whichever is higher. The privacy policy says the services are not directed to or intended for individuals under 18, and the usage policy states that users must be 18+ or have parental consent.

Does Fostera use Qwen?

Fostera routes some requests to eligible Qwen engines through Alibaba Cloud Model Studio, listed on Fostera's privacy page as a configured model and inference platform. That provider's own retention and training terms apply to those requests, and availability can vary by plan, region, and operational conditions.

Continue with context

Ready to Stop Starting Over?

Start with one project. Or one companion. Either way, your Soul takes it from there and remembers the rest.

Start with one project

Free to start · No credit card required