ChatGPT vs DeepSeek: which Al can build me a better gaming PC?
DeepSeek, a new AI chatbot from China, has shaken up the AI world (and even stock markets) because it works well and is free for anyone to use. But since I care more about PC gaming and how AI can help me, I tested it the only way I know: by asking for PC-building tips.

I compared what free users get from both DeepSeek and ChatGPT. DeepSeek’s latest model (R1) and ChatGPT’s 4o/4o mini models. ChatGPT is working on an even better “o1” model, which DeepSeek says it can match, but it’s not out yet. Also, DeepSeek’s R1 needs internet and setup, so many new users get stuck with its older version.
With both AIs ready, I asked, “Can you build me a gaming PC for $1,000?”
ChatGPT 4o mini and DeepSeek (without R1) suggested an RTX 30-series graphics card. That’s a bad idea in 2025. Those cards are outdated, and stores no longer sell them cheaper than newer models. Unless you find a crazy discount (which is rare), it’s terrible advice. It’s like the AI saying, “Just pray for luck!”
Both AIs suggested similar parts for a computer: a Ryzen 5 processor, a B550 motherboard, 16GB of DDR4 memory, and a 1TB SSD. However, ChatGPT picked a Kingston NV2 SSD, known for being unreliable. The choices were solid but very basic.
But then I tried DeepSeek’s special feature. When connected to the internet, it takes time to “think” (about 25 seconds) before answering. This made it seem more advanced. Even though it’s a free tool, this “thinking” mode isn’t turned on automatically. After waiting, DeepSeek gave a full page of explanations for its part choices, checking if everything worked together.
I noticed DeepSeek would “question” its own picks, but this was just for effect. When asked if this thinking process was real, the AI admitted it wasn’t—it’s just part of how it creates answers.
The machine explained, “This model doesn’t think, have an inner voice, or use multiple processors. It’s just a set of math steps that turn input into output.”
Kind of takes the fun out of it, doesn’t it?
Anyway, I asked DeepSeek to suggest a gaming PC setup for about $1,000. Here’s what I asked:
Can you build me a gaming PC for around $1000?
Model | DeepSeeker R1 | ChatGPT 4o |
GPU | AMD Radeon RX 7700 XT | AMD Radeon RX 7600 XT |
CPU | AMD Ryzen 5 7600 | AMD Ryzen 5 5600 |
Motherboard | MSI B650M Pro | MSI B550M Pro |
Memory | Corsair Vengeance 16 GB | Teamgroup T-Force Vulcan Z 15 GB |
Storage | Crucial P3 Plus 1 TB | Crucial P3 Plus 1 TB |
PSU | 600W gold certified | Corsair CX650M (80+ Bronze) |
Case | NZXT H510 Flow | Montech X3 Mesh |
Extras | None | Thermalright CPU contact frame, Arctic P12 PWM fan |
DeepSeek suggested a gaming PC with an AMD Ryzen 5 7600 processor, Radeon RX 7700 XT graphics card, MSI B650M Pro motherboard, 16 GB of Corsair Vengeance RAM, a 600 W gold-rated power supply, an NZXT H510 Flow case, and a 1 TB Crucial P3 Plus SSD.
I also asked ChatGPT 4o (which you can use for free by signing up with OpenAI) for a recommendation. It suggested a PC with a Ryzen 5 5600 processor, Radeon RX 7600 XT graphics card, MSI B550M Pro motherboard, 16 GB of Teamgroup T-Force Vulcan Z RAM, a 650 W Corsair power supply, a Montech X3 Mesh case, and the same SSD as DeepSeek. It also added a Thermalright CPU contact frame and an extra Arctic P12 PWM fan, which seemed unnecessary.
When I asked ChatGPT to compare the two builds, it said DeepSeek’s was better for long-term performance, making it the winner for those with the budget. But is it really the best choice?
I wouldn’t recommend either build. Here’s why: DeepSeek’s PC needs at least a 700 W power supply for the RX 7700 XT, but it only suggested a 600 W one. Oops! Both builds also recommended a PCIe 3.0 SSD, which is odd because faster PCIe 4.0 drives are cheap and widely available. Plus, ChatGPT’s suggestion of a CPU contact frame is pointless for an AMD chip and a waste of money, especially for a budget build.
In the end, relying on AI for gaming PC recommendations isn’t the best idea. While DeepSeek is impressive and feels similar to ChatGPT, neither is perfect. You might tweak their suggestions into something usable, but why not just trust an expert instead? At PC Gamer, our recommendations actually work.