FAQ

DeepSeek Navbar
FAQ - DeepSeek R1

Frequently Asked Questions

Company & Product

We are Electric Pipelines, a software engineering company in the Midwest of the United States. We specialize in consumer and business software that enables AI to do real work for regular people. If you'd like to know more about us and what we do don't hesitate to visit our website or check out some of our open source software on Github. Or, you can check us out on X (Twitter).
When DeepSeek R1 first emerged onto the scene it was met with a great deal of excitement. We were excited too, and were simply blown away with how capable R1 AI model was. When we learned that it was available for download, we set out to run it locally on our own PC's. What we found was that every available option was either too technical to set up, or required you to be a software engineer with programming tools installed on your computer just to try it. So, since we specialize in consumer software, we decided to make DeepSeek available as a consumer app that just works. Once you download and install, you open the app and it just runs: no fiddling with computer programming stuff. It runs on any Apple Silicon MacBook, Intel Mac, or Windows PC with enough memory.
DeepSeek R1 is an open source language model made by DeepSeek, a Chinese AI startup. This model is able to perform the same text-based tasks as leading models like ChatGPT, but at a lower cost.
No. We are not affiliated with the authors of the DeepSeek open source AI model. Our product provides you with the ability to run the DeepSeek R1 AI model locally on your computer. The software was written by Electric Pipelines, which is unaffiliated with the company that sponsored the open source AI model.

Purchase & Delivery

The app is a one time purchase that you download to your computer and run. There are no hidden fees. No subscription fees. Nothing more to pay. The $47 price is all you pay to be able to start chatting with DeepSeek R1 today.
After our payment processor notifies us that the payment has cleared, our system will automatically provision your personal copy of the DeepSeek R1 app, and email you a link to your copy that you can download. It usually takes just a few minutes for your link to be sent out, but we don't have control over how quickly our payment processor notifies us, so it could be a little longer. But it shouldn't take more than an hour.
Our refund policy is we will refund your purchase price if the software is defective or otherwise cannot be delivered to you. If you take delivery of the software, since it is a digital product, we do not offer a refund for a product that we've delivered.

Technical Information

This is a great question! We ship the actual DeepSeek R1 model, which is very capable. But, if you are seeing low quality answers, there's a very good reason why you might be seeing this. The app is designed to run on all computers with 4GB of memory or more. In order to deliver that kind of flexibility the app will choose an AI model that is most capable for your computer.

The quality of the answers you'll get will depend heavily on how much memory your computer has. If you have 8GB or less, the app will use the smallest, least capable, model. If you have 80GB or more, the app will use the largest and most capable model.

Model Size Free RAM Required Quality Level Capabilities and Limitations
DeepSeek-70B
80+ GB RAM
Very High
  • Near ChatGPT level performance
  • Excellent reasoning and analysis
  • Strong coding abilities
  • Consistent context retention
  • Nuanced understanding of complex topics
  • Reliable task completion
DeepSeek-32B
36+ GB RAM
High
  • Strong overall performance
  • Good reasoning and analysis
  • Capable coding assistance
  • Good context retention
  • Handles complexity well
  • Occasional minor inconsistencies
DeepSeek-14B
16+ GB RAM
Medium
  • Decent general performance
  • Basic reasoning capabilities
  • Simple coding tasks
  • May lose context in long conversations
  • Handles moderate complexity
  • Some inconsistency in responses
DeepSeek-7B
8+ GB RAM
Modest
  • Basic conversation abilities
  • Simple task completion
  • Very basic coding help
  • Limited context retention
  • May struggle with complexity
  • Frequent inconsistencies
DeepSeek-1.5B
4+ GB RAM
Low
  • Basic question answering
  • Often loses conversation context
  • May give incoherent responses
  • Limited understanding of nuance
  • Struggles with complex tasks
  • High inconsistency in quality
The DeepSeek R1 app supports Apple Silicon devices such as M1-M4, as well as Intel, and Windows.
If you cannot download the DeepSeek app on Windows because you see a message saying "DeepSeek-R1-Setup.exe isn't commonly downloaded", or your Downloads folder contains a file with a name like "Unconfirmed_216937.crdownload", don't panic! This is perfectly normal while the download is in progress.

This issue is caused by Microsoft Defender SmartScreen, which attempts to discourage anyone from opening a downloaded application until enough people download it. Microsoft does not say how many downloads is enough to eliminate this behavior.

But our app has a proper and up-to-date Authenticode Digital Signature signed by SafeNet technology company Thales. Our digital certificate is for Electric Pipelines. See Getting past Microsoft's Protection for more information.
You do not need internet to chat. The Deepseek R1 model will be downloaded to your computer after installation, and then everything will run smoothly offline.
A DeepSeek model's "intelligence" depends on its size – bigger models have more "brainpower" to handle complex information. Our DeepSeek app prioritizes privacy and runs 100% locally on your computer. To ensure a smooth, responsive experience, it automatically selects the largest model your hardware can efficiently handle.

Here's a simplified breakdown of how DeepSeek model sizes relate to their capabilities:

DeepSeek-1.5B (4+ GB RAM): Basic question answering, designed for limited resources.
DeepSeek-7B (8+ GB RAM): Better conversation, but limited complexity.
DeepSeek-14B (16+ GB RAM): Decent general performance, basic coding.
DeepSeek-32B (36+ GB RAM): Strong overall performance, good reasoning.
DeepSeek-70B (80+ GB RAM): Near ChatGPT-level, excels at complex tasks.

If performance seems lower than expected, it's likely your computer's resources are best suited for a smaller model. The app prioritizes stability and privacy over running a model that could cause slow responses or crashes. A more powerful computer (with more RAM/GPU) would allow you to run a larger, more capable model. Our app will automatically detect the best model for your Mac or Windows PC.

Privacy & Security

Yes. It runs completely on your own machine. You won't share any secrets with anyone else. The only time our app even uses the internet is to initially download the model.
All of your data and conversations are stored on your computer. The app is strictly local, and no data is ever stored anywhere else.
No one has access except for you. We do not collect your information, and your conversations cannot be recorded, even anonymously. It's all in your hands and under your control.
The software will send us a signal that it has been installed and is working correctly. We use this to determine if you have taken delivery of the software or not. After that initial notification, the software does not do any other tracking activity.

Support

The DeepSeek R1 app was designed from the ground up to just work. But, if you have any problems, you can send us an email at support@electricpipelines.com and someone will respond and help you.

More questions?

Don't see your question answered here? You can email us at support@electricpipelines.com with additional questions.