Deepseek v3.1 The New Best Cheapest Model

Feature Requests
1

Hi, I want to ask whether deepseek v3.1 will be added to cursor soon as it is just being offically annonced.

Evaluation

Category Benchmark (Metric) DeepSeek V3.1-NonThinking DeepSeek V3 0324 DeepSeek V3.1-Thinking DeepSeek R1 0528
General
MMLU-Redux (EM) 91.8 90.5 93.7 93.4
MMLU-Pro (EM) 83.7 81.2 84.8 85.0
GPQA-Diamond (Pass@1) 74.9 68.4 80.1 81.0
Humanity’s Last Exam (Pass@1) - - 15.9 17.7
Search Agent
BrowseComp - - 30.0 8.9
BrowseComp_zh - - 49.2 35.7
Humanity’s Last Exam (Python + Search) - - 29.8 24.8
SimpleQA - - 93.4 92.3
Code
LiveCodeBench (2408-2505) (Pass@1) 56.4 43.0 74.8 73.3
Codeforces-Div1 (Rating) - - 2091 1930
Aider-Polyglot (Acc.) 68.4 55.1 76.3 71.6
Code Agent
SWE Verified (Agent mode) 66.0 45.4 - 44.6
SWE-bench Multilingual (Agent mode) 54.5 29.3 - 30.5
Terminal-bench (Terminus 1 framework) 31.3 13.3 - 5.7
Math
AIME 2024 (Pass@1) 66.3 59.4 93.1 91.4
AIME 2025 (Pass@1) 49.8 51.3 88.4 87.5
HMMT 2025 (Pass@1) 33.5 29.2 84.2 79.4

Note:

  • Search agents are evaluated with our internal search framework, which uses a commercial search API + webpage filter + 128K context window. Seach agent results of R1-0528 are evaluated with a pre-defined workflow.

  • SWE-bench is evaluated with our internal code agent framework.

  • HLE is evaluated with the text-only subset.

7 Likes
2

i thought it was already in cursor for months? i see it in model list

1 Like
3

image
image370×289 11.2 KB

oh..

1 Like
4

Hey, yes, we’re already aware that this model isn’t working at the moment. I believe we’ll fix it soon.

8 Likes
5

Hi there, Cursor Team! I’d like to point out a naming inconsistency with the DeepSeek models to avoid user confusion.

  • The model currently listed as “Deepseek V3.1” (~60K context) is officially named “DeepSeek V3 0324”, as “V3.1” was an incorrect community name for it.
  • The actual “DeepSeek V3.1” was just released in August 2025. It has a 128K context length and has NOT yet been integrated into Cursor.

To fix this, could you please:

  • Rename the existing V3.1 model to its correct name: “DeepSeek V3 0324”.
  • Add the new “DeepSeek V3.1” (August 2025) model with 128K context length.
  • Update these changes in both the application and on the website to ensure consistency.

Thanks!

11 Likes
6

Nope. That one was using the wrong name by the community. The current “V3.1“ is actually “DeepSeek V3 0324”, released on March 24, 2025, with a context length of ~60K.

We’ll need to wait for the Cursor team to add the real “DeepSeek V3.1” (Aug 2025) with 128K context support.

6 Likes
7

In fact they rendered it unusable but before it was actually the ONLY really free model out there. I used it heavily for tiny questions.

Well now we want API charging, so models not bringing money is not an advantage for them. It was also changed to 0 to 0.7 if memory serves.

1 Like
8

I made this feature request 2 day earlier already and why this article upvoted than mine?:sad_but_relieved_face:

9

Hello Cursor team,

The new model DeepSeek V3.1 has just been released, and I would love to see it supported in Cursor as soon as possible.

However, I’d like to point out a potential source of confusion: the model DeepSeek V3-0324 is already listed in Cursor as “DeepSeek V3.1.” Since the newly released model is also officially named DeepSeek V3.1, this may cause ambiguity for users.

Could you please consider clarifying the naming in the model list while adding support for the new DeepSeek V3.1? That way, users will clearly understand which version they are using.

Thank you for your great work!

4 Likes
10

Hi there, Cursor Team!

I’d like to point out a naming inconsistency with the DeepSeek models to avoid user confusion.

  • The model currently listed as “Deepseek V3.1” (~60K context) is officially named “DeepSeek V3 0324”, as “V3.1” was an incorrect community name for it.
  • The actual “DeepSeek V3.1” was just released in August 2025. It has a 128K context length and has NOT yet been integrated into Cursor.

To fix this, could you please:

  • Rename the existing V3.1 model to its correct name: “DeepSeek V3 0324”.
  • Add the new “DeepSeek V3.1” (August 2025) model with 128K context length.
  • Update these changes in both the application and on the website to ensure consistency.

Thanks!

12 Likes
11

Would you please upvote this article? Doing so will help the cursor team implement this feature request more quickly.

12

Feature request for product/service

AI Models

Describe the request

pls add ds3.1 from aug 2025 version thx！！！

1 Like
13

I have a question, is the current deepseek-v3.1 v3-0324 or real v3.1? :thinking:

1 Like
14

Seems to be deepseek v3-0324, with only 60k context window size

15

image
image1084×684 41.7 KB

80K context window.

1 Like
16

That’s weird. also I voted! love the energy here.

17

Problem is that cursor has renamed v3-0324 to v3.1…
The real v3.1 was released less than a month ago.

Deepseek should have called it 3.5. or Cursor should have kept it v3-0324

18

:thinking: And cursor has not update the name by now :upside_down_face:

1 Like
19

when can i use real v3.1 in cursor?

2 Likes
20

Hey, thanks for your interest in this model. The team is considering adding an updated version of DeepSeek.

4 Likes