DeepSeek made its name in early 2025 with high-performing reasoning models that were cheap to train. Its models are published: installed on your own hardware, they analyze your campaigns and send with Mailpro, without going through the online app.
DeepSeek was founded in 2023 in Hangzhou, China. In January 2025, its R1 reasoning model surprised everyone with its results and low training cost.
Its models are strong at reasoning and coding. The full versions need a lot of hardware; lighter versions run on more modest machines.
DeepSeek's online app stores personal data in China according to its privacy policy, and several European authorities have restricted it. The usual answer from companies: install the model in-house. That is exactly what Mailpro makes possible.
| Developer | DeepSeek, Hangzhou (China) |
|---|---|
| Known for | Reasoning, coding, low training cost |
| Hardware | Full versions: a lot of hardware; light versions: more modest machines |
| Online app | Data stored in China according to its privacy policy |
| Recommended with Mailpro | The model installed on your own hardware |
Which list responds best, at what time, with which subject line: DeepSeek reasons on your real statistics.
Why this campaign underperformed: it compares and argues, step by step.
The campaign is ready; it only goes out after your approval.
A full version on a server for analysis, a light version for everyday tasks.
Install a DeepSeek model with Ollama or LM Studio, in a version that suits your hardware.
Use an interface that supports remote MCP servers: Open WebUI, LibreChat or LM Studio.
Add the Mailpro MCP server using the server address shown in your Mailpro account, on the AI Assistants (MCP) page, with OAuth sign-in (details: self-hosted AI).
In Mailpro, on the AI Assistants page, set how it sends: with approval or direct, with a daily cap.
Installed on your own hardware, DeepSeek sends nothing to its developer; only the calls to Mailpro go out, to your account in Switzerland.
The full versions need powerful servers; start with a light version for testing.
DeepSeek's recent models are published under open licenses; check the one for the version you install.
On the Mailpro side, the rules are the same for every assistant: by default, each send waits for your approval, or goes out directly within a daily cap; a log keeps track of everything, and one click suspends all sends. The details.
Installed on your own hardware, the model sends nothing to DeepSeek. Avoid the online app, however, which stores data in China according to its privacy policy.
The full versions need a lot of hardware; the light versions run on more modest machines, for simpler tasks.
No, by default every send waits for your approval in Mailpro; direct sending is set assistant by assistant.
All questions: AI assistants FAQ.
Create a free Mailpro account, connect your assistant from the AI Assistants page of your account, and stay in control of every send.
