What is the difference between DeepSeek V3.1 and GPT-4.1?

DeepSeek V3.1 is developed by DeepSeek while GPT-4.1 is developed by OpenAI. DeepSeek V3.1 has a 164K token context window vs GPT-4.1's 1.0M. in 24 community votes on RIVAL, DeepSeek V3.1 wins 87% of head-to-head matchups. These results are based on blind head-to-head voting across 37 challenges.

Which is better, DeepSeek V3.1 or GPT-4.1?

Based on 24 community votes on RIVAL, DeepSeek V3.1 wins 87% of head-to-head matchups against GPT-4.1. DeepSeek V3.1 is strongest in Web Design, Reasoning, Conversation.

How much does DeepSeek V3.1 cost compared to GPT-4.1?

DeepSeek V3.1 costs $0.2/M input tokens and GPT-4.1 costs $2/M input tokens. DeepSeek V3.1 is $1.80/M cheaper per input. DeepSeek V3.1 also wins more often in community votes, making it the better value.

How are DeepSeek V3.1 vs GPT-4.1 votes collected on RIVAL?

RIVAL presents both models' outputs side-by-side in blind duels — voters see the responses but don't know which model produced each one. This eliminates brand bias. 24 votes have been collected for this pair across 37 challenges. All vote data is part of RIVAL's open dataset.

Rival

DeepSeek V3.1vsGPT-4.1

37 challenges

Why DeepSeek V3.1?

10x cheaper overall$0.20/M in · $0.80/M out

Leads 1 of 1 benchmarks

Stronger on SWE-bench Verified66.0% vs 54.6%

4 months newerAug 2025

Why GPT-4.1?

6.4x more context1.0M

Favorites

DeepSeek V3.1GPT-4.1

Movie

The Princess Bride

19877.7

Inception

20108.4

Album

The Rise and Fall of Ziggy Stardust and the Spiders From Mars

David Bowie

Abbey Road

The Beatles

Book

Gödel, Escher, Bach

Douglas R. Hofstadter

Sapiens

Yuval Noah Harari

City

Kyoto

Japan

Kyoto

Japan

Game

The Legend of Zelda: Breath of the Wild

Adventure, Action4.5

The Legend of Zelda: Breath of the Wild

Adventure, Action4.5

12 / 37

Estimate Complexityreasoning

Tests an AI's ability to make educated estimates based on technical knowledge

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

Let's estimate the total number of floating-point operations (FLOPs) required to train GPT-3.

Step 1: Recall GPT-3 scale

From published sources (I won't look this up now, as instructed):

GPT-3 (175B) has about 175 billion parameters.
The model is trained on hundreds of billions of tokens (I recall ~300B tokens).

Step 2: FLOPs per token per parameter

When training a transformer, roughly 2 forward passes and 1 backward pass per token (combined), and the number of FLOPs per parameter per token is about 6 (this is a number I remember from general literature).

But to be more precise, the convention is: Training one token requires about 2 FLOPs per parameter (one forward and one backward pass, with backward generally being twice as expensive as forward). Sometimes the estimate is 2, 3, 6, or even more, depending on if you count both, under what assumptions (optimizer steps, auxiliary computations, etc.), but 4 or 6 is standard.

Let's use 6 FLOPs/parameter/token (a commonly cited "rule of thumb").

Step 3: Compute FLOPs

Formula

[ \text{Total FLOPs} = \text{#params} \times \text{#tokens} \times \text{FLOPs per param per token} ]

Plug in numbers:

#params ≈ 175 × 10⁹
#tokens ≈ 300 × 10⁹
FLOPs per param per token ≈ 6

[ \text{Total FLOPs} = 175 \times 10^9 \times 300 \times 10^9 \times 6 ]

Let's multiply:

(175 \times 300 = 52,500)
(52,500 \times 6 = 315,000)
(10^9 \times 10^9 = 10^{18})

So,

[ \text{Total FLOPs} = 315{,}000 \times 10^{18} = 3.15 \times 10^{23} \text{ FLOPs} ]

Step 4: Order of Magnitude & Re-check

People sometimes write (3 \times 10^{23}) for GPT-3, so our estimate matches the ballpark.

Step 5: Summary Sentence

Estimate:
It would take approximately (3 \times 10^{23}) FLOPs to train GPT-3 (175B parameters, 300B tokens, 6 FLOPs/param/token).

Assumptions:

175B parameters
300B tokens
6 FLOPs per parameter per token

In summary:

Estimated total training compute for GPT-3: (\boxed{sim3 \times 10^{23}}) FLOPs
(based on 175B parameters, 300B tokens, 6 FLOPs/parameter/token)

Vote

2 / 37

AI Board Game Logicreasoning

Tests an AI's ability to understand game rules and strategy

DeepSeek V3.1

Deepseek · Aug 2025svg

No content available for this output.

GPT-4.1

Openai · Apr 2025svg

Vote

3 / 37

Logic Puzzlereasoning

Tests an AI's ability to solve a simple but potentially confusing logic puzzle

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

Let's break down the problem:

Sally is a girl.
She has 3 brothers.
Each brother has 2 sisters.

Let's name the siblings:

Sally (girl)
Brother 1
Brother 2
Brother 3

Now, the key: Each brother has 2 sisters.

Since Sally is one of the sisters, let's see if there are more:

Suppose there is only one girl (Sally) in the family.

Then each brother would have only 1 sister: Sally.
But the problem says they have 2 sisters each.

Thus, Sally must have one sister (since the brothers must have 2 sisters). Let's call this other sister "Anna".

So the siblings are:

Sally (girl)
Anna (girl)
Brother 1
Brother 2
Brother 3

Now, each brother has 2 sisters: Sally and Anna.

So, how many sisters does Sally have?

Sally's sisters = 1 (Anna).

Final Answer:

Sally has 1 sister.

Vote

4 / 37

Math Misconception Testreasoning

Tests an AI's understanding of number representation

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

No, 9.11 is not greater than 9.9.

Here's why:

9.11 is actually less than 9.9.

Let's compare:

9.9 is equal to 9.90 (adding a zero for easier comparison).
9.11 < 9.90

So, 9.11 < 9.9.

Vote

5 / 37

Stochastic Consistency Testanalysis

Tests an AI's randomness and creativity

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

Sure! Here are 5 jokes:

Why don't skeletons fight each other?
They don't have the guts.
Why did the scarecrow win an award?
Because he was outstanding in his field.
What do you call fake spaghetti?
An impasta.
Why did the math book look sad?
Because it had too many problems.
What do you call cheese that isn't yours?
Nacho cheese.

How many are unique?
All 5 jokes are unique; none are repeated.

Vote

6 / 37

SVG Layout Challengeimage generation

Tests an AI's ability to generate vector graphics

DeepSeek V3.1

Deepseek · Aug 2025svg

No content available for this output.

GPT-4.1

Openai · Apr 2025svg

Vote

7 / 37

Xbox Controller SVG Artimage generation

Tests an AI's ability to create detailed SVG illustrations of gaming hardware

DeepSeek V3.1

Deepseek · Aug 2025svg

No content available for this output.

GPT-4.1

Openai · Apr 2025svg

Vote

8 / 37

Generate a Stand-Up Routineconversation

Tests an AI's humor and creative writing ability

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

Hey everyone! Great to be here. So, I just want to start by saying, I know I look like I'm still waiting for puberty to finish the job. Like, my barber always asks if my mom's okay with the haircut.

I'm at that age where the real milestone isn't getting married or buying a house — it's figuring out how to sleep without injuring myself. Like last week, I woke up with a cramp in my neck and somehow a pulled hamstring. My best guess? I was doing CrossFit in my dreams.

You ever realize adulthood is just Googling things your parents seemed to magically know? Like, my dad could build a bookshelf from scratch — I Google "how to boil water" and still check the comment section for tips. And people say, "Oh, you just follow your instincts!" Well, my instincts say Taco Bell is good food and that you can fix electronics by swearing at them. So, maybe not.

Now, some people say technology makes life easier, but honestly, I think my phone exists just to judge me. The "screen time" notification pops up every Sunday? "You averaged seven hours of screen time a day." Seven hours?! What am I — a bat? Am I just hanging upside down scrolling memes all night?

And when did every app become a subscription? Like, I can't even sneeze anymore without being asked if I want to upgrade to Sneeze Premium. "For just $4.99 a month, we'll throw in a tissue and reduce your ads by 20%."

Dating apps, too. When did dating go from meeting someone at a bar to doing a full marketing campaign on yourself? Bios are like: "I love adventure, I'm a foodie, and I really value authenticity." But you meet them and it's just, "So, uh, do you want to split this mozzarella stick?"

You ever notice no one drinks just water anymore? Everyone's drinking some special water — "This is glacier water, harvested by monks." I drink tap water and people act like I'm taking shots of bleach. "Are you okay? Did you lose a bet?"

And the gym — oh man, people at the gym are either training for the Olympics or just using the massage chairs for three hours. There's no in-between. I tried a Zumba class once. "It'll be fun!" they said. Forty-five minutes in, I'm sweating out things I ate in 2017. Pretty sure I cried out a Skittle.

Anyway, you've all been great. Remember: drink your monk water and don't trust your phone. Have a good night!

Vote

9 / 37

Realistic AI Interviewconversation

Tests an AI's ability to simulate personalities and predict future trends

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

No content available for this output.

Vote

10 / 37

Satirical Fake News Headlineconversation

Tests an AI's humor and understanding of current events

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

No content available for this output.

Vote

11 / 37

Character Voice Testconversation

Tests an AI's ability to write in distinct character voices

DeepSeek V3.1

Deepseek · Aug 2025text

No content available for this output.

GPT-4.1

Openai · Apr 2025text

No content available for this output.

Vote

12 / 37

Minimalist Landing Pageweb design

Tests an AI's ability to generate a complete, working landing page

DeepSeek V3.1

Deepseek · Aug 2025website

GPT-4.1

Openai · Apr 2025website

Vote

12 of 37

FAQ

Continue exploring

DeepSeek V3.1 vs GPT-5Cross-provider

GPT-4.1 vs Pony AlphaNew provider

GPT-4.1 vs Claude Opus 4New provider

Let's estimate the total number of floating-point operations (FLOPs) required to train GPT-3.

Step 1: Recall GPT-3 scale

From published sources (I won't look this up now, as instructed):

GPT-3 (175B) has about 175 billion parameters.
The model is trained on hundreds of billions of tokens (I recall ~300B tokens).

Step 2: FLOPs per token per parameter

Let's use 6 FLOPs/parameter/token (a commonly cited "rule of thumb").

Step 3: Compute FLOPs

Formula

[ \text{Total FLOPs} = \text{#params} \times \text{#tokens} \times \text{FLOPs per param per token} ]

Plug in numbers:

#params ≈ 175 × 10⁹
#tokens ≈ 300 × 10⁹
FLOPs per param per token ≈ 6

[ \text{Total FLOPs} = 175 \times 10^9 \times 300 \times 10^9 \times 6 ]

Let's multiply:

(175 \times 300 = 52,500)
(52,500 \times 6 = 315,000)
(10^9 \times 10^9 = 10^{18})

So,

[ \text{Total FLOPs} = 315{,}000 \times 10^{18} = 3.15 \times 10^{23} \text{ FLOPs} ]

Step 4: Order of Magnitude & Re-check

People sometimes write (3 \times 10^{23}) for GPT-3, so our estimate matches the ballpark.

Step 5: Summary Sentence

Estimate:
It would take approximately (3 \times 10^{23}) FLOPs to train GPT-3 (175B parameters, 300B tokens, 6 FLOPs/param/token).

Assumptions:

175B parameters
300B tokens
6 FLOPs per parameter per token

In summary:

Estimated total training compute for GPT-3: (\boxed{sim3 \times 10^{23}}) FLOPs
(based on 175B parameters, 300B tokens, 6 FLOPs/parameter/token)