Evanesca vs ChatGPT: what changes for your conversations

At a glance

The ChatGPT column describes its plans for individuals, as OpenAI's own pages read on 6 October 2026; each cell links to its page. The Evanesca column links to the page that describes it. The same comparison for Gemini, Claude, Copilot and Meta AI is on the sources page.

Where your words go, and which law applies

In the EU, you contract with OpenAI Ireland Ltd, and the law of your country of residence governs the terms source. Your conversations, though, leave Europe: OpenAI's privacy policy for Europe says it processes personal data “on servers located outside of the EEA, Switzerland and the UK”, including “in our facilities and servers in the United States”, under the Commission's standard contractual clauses where needed source.

« Content is stored on OpenAI systems and our trusted service providers’ systems in the US and around the world. »

OpenAI does offer data residency in Europe, to the customers it lists: “Eligible API customers and new ChatGPT Enterprise/Edu customers” source. The plans for individuals are not on that list. And wherever the data sits, OpenAI is a US group: the CLOUD Act lets US authorities require a provider under US jurisdiction to disclose data in its “possession, custody, or control”, wherever it is stored source.

Every model Evanesca serves runs in Europe: your words are processed in France by OVHcloud in Gravelines or Scaleway in Paris, in Mistral AI's EU region, in Germany by AKI.IO, or in Switzerland by Infomaniak. OVHcloud, Scaleway, Mistral AI and Infomaniak are European companies with no US parent, out of the CLOUD Act's direct reach; AKI.IO is a German company whose shareholders are not published. Evanesca serves no model outside Europe. Web search is off by default; if you switch it on, a search sends Linkup, a French company, only the query the model writes and the addresses of the pages it reads — never the whole conversation source. How the operators were checked.

What is kept, and what trains the next model

« Regular and archived chats remain saved to your account until you delete them or an applicable workspace retention policy removes them. »

Once you delete a chat, OpenAI schedules it for deletion “within 30 days”, unless it was already de-identified or must be kept for security or legal reasons. Temporary chats stay out of your history, but “OpenAI may keep a copy for up to 30 days for safety purposes” source.

On training, OpenAI is direct: “When you use our services for individuals, such as ChatGPT and Codex, we may use your content to train our models.” You can turn off “Improve the model for everyone” in the data controls; but a thumbs up or down can put the whole conversation it belongs to into training, even after you opt out source. Business, Enterprise and Edu workspaces are not used for training by default.

Evanesca stores no prompt and no answer: they are held in memory while the answer is written, your history lives in your browser, and of your messages only billing data — tokens and cost — is kept. Evanesca never trains a model on what you type. On the operator side, OVHcloud keeps “only the data required for billing purposes” source; Scaleway may keep requests that cause errors or look abusive for up to 14 days source; Mistral AI keeps API inputs and outputs for 30 rolling days to monitor abuse, and lets its customers object to training — Evanesca's account objects, except for what is sent to Mistral Large 4: it is a preview, and Mistral's commercial terms let it use the data of preview models for training, whatever the objection source source; Infomaniak states that queries are “neither recorded nor used to train models” source.

Who can read them

OpenAI lists four reasons for which a limited number of its staff, and service providers bound by confidentiality, may access your content: investigating abuse or a security incident, support you asked for, legal matters, and improving model performance unless you opted out source. Access is logged and limited, and the same section closes with a request:

« Please do not enter sensitive information that you would not want reviewed or used. »

On Evanesca, there is no stored copy for anyone at Evanesca to read. The operator that runs the model processes your words to write the answer; Scaleway may read requests that cause errors or look abusive, and Mistral AI keeps its models' inputs and outputs 30 days to monitor abuse. The passport under each answer says which operator it was.

Where ChatGPT is the better choice

• OpenAI's own frontier models, which Evanesca does not have; GPT-6 reasoning comes with the Plus and Pro plans source.
• More features: voice chats, scheduled tasks, custom GPTs, Codex, and an always-on agent on Pro source; Evanesca's Agent mode carries out one task at a time, when you ask.
• Apps on the web, iOS and Android, with your history saved to your account source. Evanesca runs in the browser for now and keeps history on the device you use; mobile apps and history sync are planned.
• A free plan with unlimited everyday text chats, subject to OpenAI's abuse guardrails source.
• It is running today; Evanesca has not opened yet.

For voice, OpenAI's own models, or anything that does not involve private or confidential information, ChatGPT offers more than Evanesca.

Where Evanesca fits better

• An Agent mode that carries out a task step by step — it searches the web, reads your files, writes Word, PDF and Excel documents and creates images — within a spending cap you set, and never runs code Agent mode.
• Confidential material — contracts, payslips, client files, unpublished figures — processed in Europe and not kept by Evanesca why Evanesca.
• Never used by Evanesca to train a model: not a setting to switch off, but how Evanesca works privacy.
• A passport under every answer: which model wrote it, who ran it, in which country, under which law, and what was kept passport.
• Models from several makers, each showing its checks before you use it, including OpenAI's own open-weight GPT-OSS 120B, run by Scaleway in Paris, with OVHcloud as a fallback models.
• No subscription: you pay for the answers you get, and nothing is sold or shared for advertising privacy.

Evanesca reads documents (PDF, Word, Excel, CSV) and images, draws charts, handles code and maths, and lets you change model mid-conversation. Teams get a signed Art. 28 DPA and SSO business.

Moving from ChatGPT, or using both

• Keep ChatGPT for what does not need to stay private; use the models Evanesca runs in Europe for work documents and anything you would not want reviewed.
• If you stay on ChatGPT for sensitive topics, turn off “Improve the model for everyone”, avoid the feedback buttons, and use temporary chats — which OpenAI may still keep for up to 30 days.
• Your ChatGPT history does not move to Evanesca: save what you need before you delete anything.

€2 of credit on Evanesca is enough to compare the two on real questions pricing.

How this page was checked

OpenAI's pages were read on 6 October 2026 and are quoted word for word; quotes stay in English, their original language. OpenAI updates these pages often — several were updated in the two weeks before that date — so check it. If a line is out of date, write to us and we will correct it.

• OpenAI — how OpenAI handles data in consumer services
• OpenAI — chat and file retention in ChatGPT
• OpenAI — how your data is used to improve model performance
• OpenAI — data residency and inference residency for ChatGPT
• OpenAI — Europe privacy policy
• OpenAI — Europe terms of use
• OpenAI — ChatGPT pricing

Frequently asked questions