Since its launch, it didn’t take long for DeepSeek to start ranking at number one, with many users rushing to leave positive reviews on the App Store. If you’re considering giving it a try or have been wanting to but don’t know where to start, you might be unsure how to truly experience its impressive capabilities. In this article, I’ll walk you through how to use DeepSeek effectively so you can explore its full potential and see if it’s the right tool for you.
What is DeepSeek?
DeepSeek is an advanced large language model (LLM) known for its real-time thinking and reasoning capabilities. It provides transparent and logical responses, making it stand out from other AI models like ChatGPT and Claude. DeepSeek also supports code generation and follows an open-source approach, allowing businesses to integrate and customize it according to their needs. Its rapid adoption and the creation of numerous derivative models highlight its growing impact in the AI space.
Pricing
DeepSeek R1 pricing is significantly cheaper compared to other LLMs like ChatGPT. It costs $0.55 per million input tokens and $2.19 per million output tokens.
Here’s what the pricing of DeepSeek looks like:
Model | Context Length | Max CoT Tokens | Max Output Tokens | Input Price (Cache Hit) | Input Price (Cache Miss) | Output Price |
---|---|---|---|---|---|---|
DeepSeek-Chat | 64K | - | 8K | $0.07 per 1M tokens | $0.27 per 1M tokens | $1.10 per 1M tokens |
DeepSeek-Reasoner | 64K | 32K | 8K | $0.14 per 1M tokens | $0.55 per 1M tokens | $2.19 per 1M tokens |
DeepSeek Pricing Table
How to Sign up DeepSeek?
DeepSeek is available as both a web version and a mobile app for iOS and Android, similar to ChatGPT. You can sign up using any platform, but I found the web version much easier and smoother. After signing up on the web, you can simply log into the DeepSeek app on your phone to use it on the go. Here’s how you can sign up on DeepSeek in just a few steps:
Step 1: Since we'll be using the web version, open a browser and navigate to the DeepSeek website.
Step 2: On the homepage, click the "Start Now" button to proceed.
Step 3: You’ll be redirected to the DeepSeek login page in a new tab. Since we don’t have an account yet, click "Sign Up" to create a new one.
Step 4: Enter a valid email address and set up a password for your DeepSeek account.
Step 5: Once you've filled in the required information, click "Send Code" to receive a confirmation email.
Step 6: Open your email, find the confirmation email from DeepSeek, copy the verification code, and paste it into the "Code" field.
Step 7: After filling in all the required fields, check the Terms and Conditions box and click "Sign Up" to complete the process.
Step 8: That’s it! You can now use DeepSeek for anything—whether for schoolwork, office tasks, or business needs, depending on your requirements.
If you plan to sign up using the mobile app, the steps are exactly the same. You'll need to fill in the same details and confirm your email. Unlike ChatGPT, DeepSeek doesn’t require you to link a mobile number—just an email is enough.
How to Use DeepSeek?
With your account now created on DeepSeek, let’s explore how to use the web version for day-to-day tasks. It’s quite similar to ChatGPT, so the learning curve is minimal. Whether you’re working on research, writing, or problem-solving, you’ll find the interface intuitive and easy to navigate.
Step 1: Once you sign up and create your account, you'll be able to access DeepSeek Chat, where you can use the chatbox to enter your prompts and queries to get answers from DeepSeek.
Let’s take a look at an example of using the default DeepSeek version to write a short story.
Example Prompt:
"Write a short horror story about a detective who discovers a hidden underground city".
Step 2: By default, DeepSeek’s Chat version is selected, but for more advanced reasoning, you can click on the "DeepThink (R1)" option to switch to the DeepSeek Reasoning model.
The DeepSeek Reasoning model is an advanced version that critically analyzes the data within the prompt to provide more logical and accurate answers. You can use the reasoning model in many scenarios, such as:
Math Problem Example:
"If a train travels at 80 km/h for 3 hours and then increases its speed by 50% for the next 2 hours, how far does it travel in total"?
Problem-Solving and Math Reasoning Example:
"You have 8 red and 6 blue marbles in a bag. If you randomly pick two marbles, what is the probability that both are red?"
Coding Assistance Example:
"Write a Python function that takes a list of numbers and returns the sum of all even numbers".
Writing & Creativity Example:
"Write a 4-line motivational poem about overcoming challenges".
Research and Summarization Example:
"Summarize the key points of Einstein’s theory of relativity in simple terms".
Step 3: There may be times when you feel the responses aren't quite right, which could be because the prompt wasn't clear enough. In such cases, you can always reply with a follow-up prompt like:
"Explain that in simpler terms."
Step 3: You can also click on the "Search" button to activate online search, allowing DeepSeek to fetch answers from the internet for the most up-to-date information.
I found the DeepSeek AI research really helpful, and one main reason for this is that when we search on search engines like Google, we don’t have the luxury of explaining exactly what we’re looking for. However, with the DeepSeek search feature, I could type my queries in more detail and ask DeepSeek to go through the search engine to find the most relevant data for my needs.
DeepSeek Search Example:
“I’m looking for research papers focusing on Scope 3 sustainability within the confectionery or FMCG industry, published after 2015. The papers should be from reputable, high-quality journals.”
How to Use DeepSeek Locally?
AI relies heavily on GPU power, which is why GPU prices have surged since the rise of AI applications. Large language models (LLMs) like DeepSeek and ChatGPT also require GPUs to function efficiently. However, DeepSeek R1 offers the advantage of running locally without depending on cloud-based AI servers.
DeepSeek R1 includes a range of text-based models, along with a 70B Vision variant for image analysis. For general use, you can run models up to 14B parameters without a GPU. This allows for tasks such as writing, chatting, coding, deeper reasoning, and research.
Before getting into the steps for running DeepSeek locally, let’s first go over the system requirements:
Model | RAM Required | CPU Required | GPU Needed? | Best Use Cases |
---|---|---|---|---|
1.5B | 8GB+ | Any modern CPU | No | Basic writing, chat, quick responses |
8B | 16GB+ | 4+ Cores (Intel i5/Ryzen 5/M1) | No | General reasoning, longer writing, coding |
14B | 32GB+ | 6+ Cores (Intel i7/Ryzen 7/M2) | No | Deeper reasoning, coding, research |
With the right hardware, DeepSeek runs entirely offline—no API calls, third-party logging, or cloud dependencies. This ensures complete privacy while giving you a personal LLM assistant under your control.
Download Ollama on Your System
To run DeepSeek locally, we will be using Ollama, a free software that allows you to run LLM models offline. Here’s how you can download and install Ollama on your system:
Step 1: To run DeepSeek, you first need to download Ollama, a software that allows you to run LLM locally. Simply open your browser, go to the Ollama website, and click on "Download for Windows" to begin the download.
Step 2: Once the download is complete, open the Ollama installer file to launch the installation wizard.
Step 3: The Ollama Setup window will open. Click on "Install" to continue.
Step 4: The installation may take some time, but once completed, Ollama will start running in the background.
Run PowerShell
Once Ollama is installed, the next step is to run PowerShell, where we will use a command to run DeepSeek locally on our system.
Step 1: To open PowerShell, search for "PowerShell" in the search bar on the taskbar and click on "Windows PowerShell" to open it.
Now, you need to copy the command to run DeepSeek on your system. To do this, you will need to go back to the Ollama website.
Step 2: On the Ollama website, click on the "Models" tab in the toolbar and select "deepseek-r1" from the list of available LLM models.
Step 3: Click on the "Parameters" dropdown and select the model according to your PC specifications.
Note: Ensure you select the right model; otherwise, your experience using DeepSeek locally may be very slow due to system limitations.
Step 4: Once you have selected the parameters, click on the "Copy" icon to copy the Ollama command.
Now that we have copied the Ollama command to run DeepSeek on our system with the correct parameters, let’s go back to Windows PowerShell.
Step 5: In Windows PowerShell, press “CTRL + V” to paste the command we copied, then press “Enter” to begin downloading DeepSeek on your system.
Note: The download may take several minutes, depending on the DeepSeek version and your internet speed. Higher versions mean larger files.
Step 6: Once DeepSeek is downloaded locally, you can start interacting with it. Simply type your message below the download completion message.
Step 7: You can now interact with DeepSeek locally. To test it, disconnect from the internet and type a message.
After following all these steps and exploring the different methods DeepSeek offers, you should be able to run it smoothly and integrate it into your workflow efficiently. Whether you're using it for research, data analysis, or problem-solving, DeepSeek's deep-thinking capabilities will help you uncover insights beyond the surface. The more you experiment with its features, the better you'll understand how it can cater to your specific needs.
FAQs
1. What is the difference between DeepSeek and ChatGPT?
DeepSeek is superior for technical tasks such as coding and debugging, offering advanced customization while being open-source and free to use. In contrast, ChatGPT excels in user experience, structured data presentation, and privacy compliance, making it better suited for general conversations. The best option varies based on your individual needs.
2. Is it safe to use DeepSeek?
DeepSeek's advanced capabilities and customization options make it a powerful tool, but concerns about data privacy and security remain. Users should be mindful of the information they share while using the platform. To use DeepSeek AI safely, follow these precautions:
Avoid sharing sensitive data like your name, address, or confidential details.
Verify AI-generated information to prevent misinformation.
Be cautious in work settings by adhering to company security policies and avoiding AI-assisted automation without human oversight.
Refrain from using DeepSeek on work devices to prevent data exposure.
Use a VPN to enhance privacy by masking your IP and encrypting internet traffic.
3. What can I do if I fail to sign in?
There can be instances where you might not be able to sign in to your account. This could be due to DeepSeek's strict security measures temporarily restricting access based on recent activity. However, to resolve this, you can submit an appeal through the provided form and wait for a DeepSeek representative to contact you and clarify the issue.
Uncovering Insights with DeepSeek
The best part about DeepSeek, in my opinion, is its deep-thinking ability. I love how it provides the backstory behind conclusions, so the results make more sense. It’s like solving a math problem—not just getting the answer, but also having the concepts cleared up. The tool helps users understand the "why" behind the "what", offering clarity and deeper comprehension. You should definitely try it out to see how it caters to your specific use cases and helps solve your problems, whether it's research or data analysis.