Deepseek has caused quite a stir since its release in January, and is considered a direct competitor against Open AI’s already popular chat GPT. But what is the difference between Chat GPT and Deepseek and which platform should you use? Reportedly, Deepseek is as equally powerful as chat GPT, able to do coding, mathematics and more, but is there anything else we should know? 

Appearance

Appearance-wise, the two platforms initially look quite similar. You can upload images to both,  with similar colour palettes, and both platforms bear a striking resemblance to each other.  

 

 

 

 

The prompt

I gave chat GPT and Deepseek the same prompt and here are their responses below. When typing the command out, Chat GPT did not correct any spelling mistakes whereas Deepseek corrected the mistakes automatically as I typed them, a small but notable feature. 

 

“Can you please make this copy better: SEO is a crucial part of marketing and website optimisation. Without SEO you are severely limiting your website’s capabilities and reach.”

Chat GPT’s Response:

Certainly! Here’s a stronger and more polished version:

SEO is essential for effective marketing and website optimisation. Without it, you’re limiting your website’s potential, making it harder for people to find and engage with your content.

Deepseeks response:

Certainly! Here’s a more polished and engaging version of your copy:

“SEO is an essential component of modern marketing and website optimisation. Without it, you’re not just limiting your website’s potential—you’re missing out on reaching a broader audience and maximising your online visibility. Investing in SEO ensures your site stands out in a crowded digital landscape.”

Let me know if you’d like further tweaks!

 

Again – the similarities are clear down to the precise wording, keeping the beginning of the text similar to the one provided, Deepseek branched out more and provided a much lengthier answer in a bid to be ‘engaging’. In my opinion, it is a little too long but this is easily rectified and sounds arguably more ‘human’ than chat GPTs response.  

Notable mentions

Recently I came across a podcast episode by David Perell featuring Tyler Cowen titled ‘The Most Practical Conversation About Writing With AI’. Cowen makes some great points about the versatility of Deepseek from a content writing perspective. Explaining that Deepseek is less manipulated to sound a certain way. “It’s less bland, I would say its better at poetry, better at emotion, more romantic, more uneven”. He does note that it ‘hallucinates more’ however if you are aware of this and fact check it, Deepseek can quickly become a great tool to assist your copywriting.

During the podcast, there is a heavy emphasis on continuous learning and adaptation.  Perell and Cowen repeatedly discuss the necessity for people to keep evolving and learning more about AI. Criticality and the need for a ‘human touch’ is also greatly emphasised. AI can be flawed and produce flawed results, and it is not a replacement for the human mind, simply an assistant or a tool to enhance our work. 

The technical differences between the two platforms

Energy consumption.

One huge difference between Chat GPT and Deepseek, is their energy usage and sustainability. Deepseek uses 20-30% less energy than Chat GPT, making it hugely more environmentally friendly compared to Chat GPT. However, this does not mean that Deepseek is sustainable. It can be argued that AI will never be sustainable due to a large amount of energy it consumes, but energy saving where possible is still incredibly important. 

Cost.

Chat GPT runs at a high operational cost, due to the high-end graphics processors they use for their AI,  which require significant cooling. DeepSeek claims to have found a workaround for this. Using more efficient hardware and custom accelerators designed specifically for its AI model Deepseek was able to make their AI less costly to run. 

For users, Deepseek is completely free, whereas chat GPT has a free model and their premium package costs 20 dollars a month. The free answers on Chat GPT are arguably more robotic than the paid verison, which is similar to the wording used by Deepseek. Yet with a little tweaking of how you phrase the question, you can achieve very similar results on Chat GPT’s free version.

Training the platforms

It’s reported that Chat GPT spent billions on their training for the programme. Deepseek’s training however reportedly only cost 5.6 million, as they took a different approach using fewer AI accelerators. 

Open Ai’s models are trained on 3 main sources of information:

1. Publicly available information on the internet
2. Information from third party partners
3. Information that users or ‘human trainers’ provide to the platform or generate

 

Deepseek uses similar methods, but also uses reinforcement learning which has played a huge role in advancing their language models’ reasoning capabilities. 

Word filtering

As far as I am aware, Chat GPT doesn’t filter or hide any words, whereas Deepseek as a Chinese-owned company does contain some content filtering. However, it seems that most censoring can be easily avoided by simply not using the app and using the browser version. Sadly,  there are still certain biases established during training cannot be removed as easily. 

What data do the platforms collect and store?

When asked what data they collect, here were the two platforms responses:

Deepseek:

Chat GPT:

 

To summarise

Chat GPT and Deepseek are unarguably similar. There are no real stand out differences to place one ahead of the other, and use simply comes down to personal preference. The general consensus across the internet is that a healthy mix of both (and even Claude AI for explanations and text breakdown) is the best way to get what you want out of AI. Deepseek seems to be better at mathematics, coding and algorithms where as Chat GPT has no content filters and will give you a candid recap of certain historic events that Deepseek cannot.

Personally, I’ll stick to ChatGPT for more data-driven tasks but experiment with Deepseek for copywriting queries or situations where ChatGPT’s depth feels unnecessary. For more Chat GPT comparisons, see our Gemini v ChatGPT blog.