3

I have been using ChatGpt for clear my confusion related to different courses in my semester. But currently another LLM model DeepSeek-R1 is launched.

My question is which model has higher accuracy than other, ChatGpt or DeepSeek?

Also there are another model that is related to DeepSeek is Qwen DeepSeek, which model has higher accuracy than other, DeepSeek-R1 or Qwen DeepSeek?

nbro
  • 42,615
  • 12
  • 119
  • 217
Monte_carlo
  • 41
  • 6

2 Answers2

8

First of all, ChatGPT is not just one model. It's the tool you can interact with and which uses a model under the hood. The tool and models have evolved over the years (and this will continue to happen) and you can switch between models or pay for better models. The same or similar with other tools, like Gemini and likely DeepSeek.

Second, these models have been evaluated on different tasks and they don't perform the best on all of them, so we cannot say that one is better than the other in all cases.

Third, the actual DeepSeek website has a graph that illustrates how DeepSeek R1s compare with other models.

Moreover, DeepSeek V3 and R1s are not the same thing.

The DeepSeek V3 and R1s papers probably contain more info.

nbro
  • 42,615
  • 12
  • 119
  • 217
0

Also depends on the kind of question. Some questions are probably not in the standards tests but that are asked by real users.

Ask chatGPT (whatever version) and DeepSeek (whatevers version) about politics in China, human rights and so on. The answers will ... hum .. differ.

Arthur Weasley says that, when some magic object seems to think by itself, you always have to know where the brain is.

EDIT : this answer is about DeepSeek as publicated, but not a fundamental characteristic of the training mechanisms described in the papers. The openness and the low cost of DeepSeek allows more or less everybody to train its own model with its its own biases.

Laurent Claessens
  • 101
  • 2