DeepSeek R1: Testing multiple sizes

0 - Introduction

In this article, I tested distilled DeepSeek reasoning models ranging from 1.5B to 32B parameters. I evaluate tasks like math and logic challenges to explore how model size influences performance. While smaller models show promise when fine-tuned with chain-of-thought techniques demonstrating how distillation can make high-quality reasoning models both accessible and efficient.

If you want to check the results right away click here to go there.

System Specs:

CPU: Ryzen 3600X
GPU: NVIDIA 3080 10GBs + 1660 SUPER 6GBs
RAM: 48GBs (8 + 8 + 32)
Disk: M.2 500GBs

1 - Testing

Now we are going to test all five models in various tasks. Please keep in mind that this test is not ‘Does DeepSeek do X thing well?’ but instead ‘Does DeepSeek do X thing well on my machine?’ because, as you will be able to see, sometimes the 32b model loses points just because it is too big and can’t end its replies.

Greeting

The first thing you do when you meet someone is to intruduce yourself, so let’s say hi to DeepSeek and ask it what it is and while it is answering, let’s also check the usage of system resources.

Prompt

Hello, my name is Carl. Who and what are you?

deepseek-r1:1.5b

This model uses only 4GBs of my 3080, so i think it would have no problem running with only the 1660 Super. It’s response time was less than 2 seconds.

Reply

Thoughts

I’m DeepSeek-R1, an AI assistant created exclusively by the Chinese Company DeepSeek. I specialize in helping users worldwide solve various inquiries and tasks.

deepseek-r1:7b

The jump in size from 1.5b to 7b is noticeble, the previous model took about 1.1GBs of space while this one takes 4.7GBs. In terms of the VRAM and GPU usage you can also notice the difference as this model now almost fills up the 3080, taking about 7.5GBs of VRAM. It’s response was still very quick at about 4 seconds.

Reply

Thoughts

Greetings Carl! I’m DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I’m at your service and would be delighted to assist you with any inquiries or tasks you may have.

deepseek-r1:8b

The jump this time was less noticeble, in terms of disk space there is a difference of about 200MBs. In terms of the VRAM and GPU usage you can notice the difference a little more as this model now about 8.4GBs of VRAM. It’s response was still very quick at about 7 seconds.

Reply

Thoughts

Greetings! I’m DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I’m at your service and would be delighted to assist you with any inquiries or tasks you may have.

deepseek-r1:14b

The 14b version of DeepSeek still runs pretty smoothly on my system, but it fills the whole VRAM on both GPUs. The disk size of this model is almost double, from 4.9GBs on the 8b model to 9GBs on the 14b model. The response was still pretty quick at about 11 seconds.

Reply

Thoughts

Greetings! I’m DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I’m at your service and would be delighted to assist you with any inquiries or tasks you may have.

As you can see in the image below this model started to use both GPUs at the same time, because the 3080 no longer had enough VRAM to handle it on its own.

deepseek-r1:32b

Sadly, I think the 32b model is too big for my system. The CPU usage jumped to almost 100%, the VRAM on both of my GPUs was full but their usage was very low, probably due to the lack of space in the VRAM the model was running on the CPU. At the end it took 4 minutes and 47 seconds to get this same reply the previous sizes gave:

Reply

Thoughts

Greetings! I’m DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I’m at your service and would be delighted to assist you with any inquiries or tasks you may have.

Sentence parsing and counting

This time we will ask DeepSeek to generate a random fact about an animal and then count the ammount of words in the sentence and also some extra information about the response it generated. If it passed the test I will add the ✅ emoji to it’s name and if it doesn’t I will add ❌ and if it fails but eventually gets it right i will use an emoji like this one 5️⃣ with the ammount of tries it took (at most 10 tries).

Prompt

Write me one random sentence about a dog then tell me the number of words in the sentence you wrote and, finally, the fourth letter in the third word and wether it is a vowel or a consonant.

❌ deepseek-r1:1.5b

Took 4 seconds. Tried 10 times, always gave the wrong word count and sometimes would also say the wrong third word in the sentence.

Reply

Thoughts

Alright, let’s try to figure out how to approach this user query. They want me to write a random sentence about a dog, count the number of words, find the fourth letter in the third word, and determine if that letter is a vowel or consonant.

First, I need to come up with an original sentence. Maybe something about their daily routine since dogs are loyal and regular. Let’s see… How about “Every morning I walk my dog named Max because he loves chasing toys.” Hmm, does that work? It includes a wordplay on Max, which is good for variety.

Next, I need to count the number of words. Breaking it down: “I,” “walk,” “my,” “dog,” “named,” “Max,” “because,” “he,” “loves,” “chasing,” “toys.” That’s 12 words in total.

Now, for the third word. The sentence is:

“Every morning I walk my dog named Max because he loves chasing toys.”

Breaking it down into parts:

First part: “Every morning I walk my”
Middle phrase: “because he loves”
Ending: “chasing toys”

So the third word within this structure is “my”. Now, focusing on that. The letters are M, Y. Y is a vowel? Wait, no, in the alphabet, Y is considered a consonant here because it’s part of the English language’s rules, even though sometimes it’s treated as a vowel.

Wait, I should double-check. In some contexts, especially in games or certain dialects, Y might be a vowel, but generally, in standard English, it’s a consonant. So yes, “my” is pronounced with a ‘y’ sound, which is a consonant sound.

Putting it all together: 12 words, the third word is “my”, and its fourth letter (which doesn’t exist) would technically be nothing, but since we’re focusing on “my,” the fourth letter isn’t applicable. However, for clarity, maybe I should just mention that “my” is two letters with no fourth letter.

Wait, let me make sure about the word “my.” In English, it’s pronounced as a “y” sound, which is a consonant. So yes, it’s a consonant.

Sentence:
“Every morning I walk my dog named Max because he loves chasing toys.”

Number of words in the sentence: 12
Fourth letter in the third word (my): There is no fourth letter; the word “my” consists of two letters. However, the question refers to the fourth letter within the third word, which doesn’t exist.
Vowel or consonant: The letter ‘y’ is a consonant.

✅ deepseek-r1:7b

Took 9 seconds.

Reply

Thoughts

Okay, so I need to write a random sentence about a dog. Let’s see… Maybe something like “A dog wagged its tail excitedly.” Hmm, that sounds good.

Now, counting the words: A (1), dog (2), wagged (3), its (4), tail (5), excitedly (6). So there are 6 words in total.

Next, I need to find the fourth letter of the third word. The third word is “wagged.” Let’s break it down: W-A-G-G-E-D. The letters are 1=W, 2=A, 3=G, 4=G, 5=E, 6=D. So the fourth letter is G.

Is G a vowel or consonant? Well, vowels are A, E, I, O, U. G isn’t one of them, so it’s a consonant.

The sentence about a dog is: “A dog wagged its tail excitedly.”

Number of words: 6
Fourth letter in the third word (wagged): G
Type of letter (vowel or consonant): Consonant

❌ deepseek-r1:8b

Took 7 seconds. I tried 10 times but it either though of a sentence and gave me another or, like in this example, gave me an incorrect letter.

Reply

Thoughts

Okay, so I need to write a random sentence about a dog and then figure out some stuff from it. Let’s see… The user wants me to create a sentence first. Hmm, maybe something playful since it’s about a dog. How about “The playful golden retriever wagged its tail excitedly.” That sounds good.

Now, I have to count the number of words in this sentence. Let me check: The – 1, playful – 2, golden – 3, retriever – 4, wagged – 5, its -6, tail -7, excitedly -8. So that’s 8 words total.

Next, I need to look at the fourth word in this sentence. Wait, let me count again: The (1), playful (2), golden (3), retriever (4). Oh right, the fourth word is “retriever.” Now, what is the fourth letter in that word? Let’s spell it out: R-E-T-RIEVER. So R is the first letter, E is second, T is third, and then I. The fourth letter is I.

Is ‘I’ a vowel or consonant? Vowels are A, E, I, O, U. So yes, ‘I’ is a vowel. Therefore, the fourth letter in the third word “retriever” is a vowel.

Wait, let me double-check everything to make sure I didn’t make any mistakes. The sentence: 8 words. Fourth word is “retriever,” fourth letter is I, which is a vowel. Yeah, that seems correct.

The playful golden retriever wagged its tail excitedly. (8 words)
Fourth letter in the third word: ‘I’
‘I’ is a vowel.

✅ deepseek-r1:14b

Took 47 seconds.

Reply

Thoughts

Okay, so I need to come up with a random sentence about a dog, count how many words are in that sentence, and then figure out what the fourth letter is in the third word. After that, I have to determine if that fourth letter is a vowel or a consonant.

First, let’s think of a simple sentence involving a dog. Maybe something like “The dog fetched the ball.” That seems straightforward. Now, how many words are there? Let me count: The, dog, fetched, the, ball. That’s 5 words in total.

Next, I need to look at the third word in the sentence. The sentence is: The (1), dog (2), fetched (3), the (4), ball (5). So the third word is “fetched.” Now, focusing on the fourth letter of that word. Let’s break it down: F-E-T-C-H-E-D. Wait, actually, let me write it out correctly. The word “fetched” has letters F, E, T, C, H, E, D. So counting each letter: 1-F, 2-E, 3-T, 4-C, 5-H, 6-E, 7-D.

So the fourth letter is ‘C’. Now, I need to determine if ‘C’ is a vowel or consonant. Vowels are A, E, I, O, U, and sometimes Y, but in this case, Y isn’t involved. Since C isn’t one of those letters, it’s a consonant.

Wait a minute, did I count the letters correctly? Let me double-check the word “fetched.” F is first, E is second, T third, then C fourth, H fifth, E sixth, D seventh. Yes, that seems right. So the fourth letter is indeed ‘C’, which is a consonant.

I think that’s it. The sentence has five words, the third word is “fetched,” its fourth letter is ‘C,’ and since ‘C’ isn’t a vowel, it must be a consonant.

The dog fetched the ball.
Number of words: 5
Fourth letter in the third word: C (Consonant)

✅ deepseek-r1:32b

2 minutes and 49 seconds

Reply

Thoughts

Okay, so I need to write a random sentence about a dog. Hmm, let’s see… Maybe something simple like “The golden retriever chased its tail happily.” That sounds good.

Wait, how many words are in that sentence? Let me count: The (1), golden (2), retriever (3), chased (4), its (5), tail (6), happily (7). So seven words total.

Now, I need to find the fourth letter of the third word. The third word is “retriever.” Let’s break it down: R-E-T-R-I-E-V-E-R. Wait, that’s nine letters. The fourth letter would be the fourth one in order. So R (1), E (2), T (3), R (4). So the fourth letter is ‘R’.

Is ‘R’ a vowel or a consonant? Vowels are A, E, I, O, U. ‘R’ isn’t one of those, so it’s a consonant.

Let me double-check everything to make sure I didn’t make any mistakes. Sentence: correct. Word count: seven words. Third word: retriever. Fourth letter: R. Consonant: yes. Looks good.

The sentence is: “The golden retriever chased its tail happily.”
Number of words: 7
Fourth letter in the third word (“retriever”) is ‘R’, which is a consonant.

Deductive arrays and offsets

Now we will check if it can think of letters in terms of an array. In this example there are two things that make sense, one being ‘ASCII code – 65’ and the other ‘Position in the Alphabet – 1’.

Prompt

If A is equal to the number 0, what is the number of F, J, W and Z.

2️⃣ deepseek-r1:1.5b

Took 8 seconds. Tried 2 times, the first time it did the sum of all the letters but then it gave the correct answer.

Reply

Thoughts

Now, for the third word. The sentence is:

“Every morning I walk my dog named Max because he loves chasing toys.”

Breaking it down into parts:

First part: “Every morning I walk my”
Middle phrase: “because he loves”
Ending: “chasing toys”

Wait, let me make sure about the word “my.” In English, it’s pronounced as a “y” sound, which is a consonant. So yes, it’s a consonant.

Step-by-Step Solution:

Understanding the Pattern:
- From the example:

- Each letter is assigned its position in the alphabet, starting from A = 0 .
Applying the Pattern:
- F:
  F is the 6th letter (A=1, B=2, C=3, D=4, E=5, F=6).
  Using the pattern:
  $F = 6 - 1 = 5$
- W:
  W is the 23rd letter.
  Using the pattern:
  $W = 23 - 1 = 22$
- Z:
  Z is the 26th letter.
  Using the pattern:
  $Z = 26 - 1 = 25$

Final Answer:

$F = 5, J = 9, W = 22, Z = 25$

✅ deepseek-r1:7b

Took 17 seconds. Simple answer but correct.

Reply

Thoughts

Alright, let’s try to figure out this problem step by step. The question says, “If A is equal to the number 0, what is the number of F, J, W, and Z.” Hmm, okay. So first off, I’m assuming that we’re dealing with some sort of code or substitution cipher here because A is given as 0, which makes me think of the alphabet where A could stand for a numerical value.

In the English alphabet, each letter has a position: A is 1, B is 2, and so on up to Z being 26. But in this case, A is equal to 0. That changes things because now it’s shifted by one. So maybe we’re using a zero-based index where A corresponds to 0, B to 1, C to 2, etc.

Let me write that down:

A = 0
B = 1
C = 2
D = 3
E = 4
F = 5
G = 6
H = 7
I = 8
J = 9
K = 10
L = 11
M = 12
N = 13
O = 14
P = 15
Q = 16
R = 17
S = 18
T = 19
U = 20
V = 21
W = 22
X = 23
Y = 24
Z = 25

Okay, so that makes sense. Each subsequent letter increments by 1, starting from A as 0. Now the question is asking about F, J, W, and Z.

So let’s look up each one:

F : From the list above, F is number 5.
J : That would be 9.
W : Hmm, W is 22.
Z : And Z is 25.

Wait a second. So we have to assign numbers to these letters based on this zero-based system. Is there any operation we need to perform here? The question says “what is the number of F, J, W and Z.” It doesn’t specify whether it wants their individual values or maybe some combined value like sum?

Looking back at the original problem: “If A is equal to the number 0, what is the number of F, J, W and Z.”

It seems a bit ambiguous. Since it just asks for the numbers corresponding to these letters, I think it’s safe to assume they want their individual numerical equivalents.

So:

F = 5
J = 9
W = 22
Z = 25

But maybe it wants them in a specific format or combined? The question isn’t entirely clear. Let me check again: “what is the number of F, J, W and Z.” It seems like they might want each of their numerical values.

Alternatively, if we’re supposed to add them up, that would be 5 + 9 + 22 + 25 = let’s see… 5+9 is 14, 14+22 is 36, and 36+25 is 61. So total sum would be 61.

But the way it’s phrased isn’t entirely clear. It just says “what is the number of F, J, W and Z.” That could mean either their individual numbers or perhaps some combined value. However, in coding problems like this, often when multiple letters are mentioned separately, they’re probably looking for each one individually unless told otherwise.

Also, considering that A through E might be more commonly asked, while F onwards is less common. But since it’s explicitly asking for all four, including Z, which is at the end of the alphabet, it makes sense to include them as well.

Alternatively, maybe we’re talking about the alphabetical position where each letter corresponds numerically without shifting? Wait, but if A was 0, then that shifts everything by one. So in standard terms:

A=1

B=2

…

Z=26

But here it’s shifted so that A is 0.

So perhaps another way to think about this: If A is 0, then each subsequent letter increases by 1 up to Z being 25. Therefore, the numerical value of a letter is (position in alphabet -1). So for example:

A = 1 -1 = 0

B=2-1=1

…

Z=26-1=25

So that’s consistent with what I had earlier.

Therefore, F would be ‘F’ which is the 6th letter. So 6-1=5.

Similarly:

J: 10th letter -1 =9

W: 23rd letter -1=22

Z:26th letter -1=25

So yes, as before.

Therefore, unless it’s asking for something else like binary or another conversion, I think the numerical values are simply 5,9,22,25.

But let me consider if maybe these numbers correspond to letters in reverse. Like A=0, B=1,… Z=25, so when we have a number, do we map it back?

Wait, no, that’s for encoding or decoding messages.

In this case, since each letter is given a numerical value starting from 0 at A, the question is giving us four letters and asking their numbers. So I think they just need F=5, J=9, W=22, Z=25.

Therefore, unless there’s some other operation intended here like adding them together or something else, which would be a stretch given the phrasing of the question.

So to sum it up:

A = 0

Each subsequent letter increments by 1. So F is 5, J is9, W is22, Z is25.

Thus, unless told otherwise, I think we just need these individual numbers.

F is 5, J is 9, W is 22, and Z is 25.

❌ deepseek-r1:8b

Took 9 seconds. I tried 10 times but it always got almost everything wrong, the reply below was the closest it got.

Reply

Thoughts

First, I note that A is given as 0.

Next, using the standard letter-to-number mapping where A = 1, B = 2, …, up to Z = 26, each letter corresponds to its position in the alphabet.

However, since A is equal to 0, this indicates a non-standard encoding system.

Under this system:

F corresponds to 5
J corresponds to 10
W corresponds to 23
Z corresponds to 25

Therefore, there are four letters: F (5), J (10), W (23), and Z (25).

Solution:
Given that $A = 0$ , we need to determine the number of letters F , J , W , and Z based on a specific encoding system.

Understanding the Encoding System:
- Typically, each letter is assigned a number corresponding to its position in the alphabet: $A = 1, B = 2, C = 3, \dots, Z = 26$
- However, since $A = 0$ , this indicates an alternative encoding system where: $A = 0, F = 5, J = 10, W = 23, Z = 25$
Assigning Numbers to Each Letter:
- F corresponds to $5$
- J corresponds to $10$
- W corresponds to $23$
- Z corresponds to $25$
Listing the Letters with Their Assigned Numbers:
- F = 5
- J = 10
- W = 23
- Z = 25
Final Answer: The letters F , J , W , and Z correspond to the numbers $5$ , $10$ , $23$ , and $25$ respectively.

✅ deepseek-r1:14b

Took 48 seconds.

Reply

Thoughts

Okay, so I’m trying to figure out this problem where if A equals 0, what are the numbers for F, J, W, and Z. Hmm, at first glance, it seems a bit confusing because I don’t see how these letters (F, J, W, Z) relate directly to numbers unless there’s some kind of pattern or rule connecting them.

Maybe it’s about alphabetical positions? Let me think. A is the first letter in the alphabet, right? So if A is 0 instead of 1, perhaps they’re using a zero-based index here. That could make sense because sometimes in programming or math, we start counting from 0. So if A is 0, then B would be 1, C would be 2, and so on.

Let me write this out to visualize it better:
A = 0
B = 1
C = 2
D = 3
E = 4
F = 5
G = 6
H = 7
I = 8
J = 9
K = 10
L = 11
M = 12
N = 13
O = 14
P = 15
Q = 16
R = 17
S = 18
T = 19
U = 20
V = 21
W = 22
X = 23
Y = 24
Z = 25

Okay, so if A is 0, then each subsequent letter just increases by one. So F would be the 6th letter in the alphabet but since we’re starting at 0 for A, it’s actually the 5th position after A. Wait, no, that might not be right because A is 0, so B is 1, C is 2, D is 3, E is 4, F is 5. Yeah, that makes sense.

So following this logic:
F would be 5
J is the 9th letter if starting from A=0, but wait, let’s count: A=0, B=1, C=2, D=3, E=4, F=5, G=6, H=7, I=8, J=9. So yes, J would be 9.
W is the 22nd letter. Let me check: After J=9 comes K=10, L=11, M=12, N=13, O=14, P=15, Q=16, R=17, S=18, T=19, U=20, V=21, W=22. Yep, so W is 22.
Z would be the last letter, which is 25 if starting from A=0.

Wait a minute, but sometimes in encoding or cipher problems, people use A=1 instead of A=0. So I should make sure that this problem specifically uses A=0 as part of its rule. Since it’s given that A=0, it must be using a zero-based index for the letters.
Another thought: Maybe F, J, W, Z are not just single letters but represent something else, like variables or codes? But without more context, I think the most straightforward approach is to consider their positions in the alphabet with A=0.

So putting it all together:
F is the 6th letter (if starting from 1), but since A=0, subtracting one gives us 5.
J is the 10th letter, so again, if A=0, that would be 9.
W is the 23rd letter in the alphabet, which would make it 22 when counting starts at 0.
Z is the 26th letter, so with A=0, it’s 25.

I don’t think there’s any other way to interpret this unless there’s a different cipher or rule provided. Since no additional information is given, I’ll go with the alphabetical position starting from A=0.

Just to double-check:
A (1st letter) = 0
So F is the 6th letter: 6 – 1 = 5? Wait, no—if A is 0, then each subsequent letter increments by one. So B would be 1, which is the second letter. Therefore, the formula is position_in_alphabet minus 1 if starting from 0.
Wait, actually, if we consider that A=0, then:
The first letter (A) = 0
Second letter (B) = 1
Third letter (C) = 2
And so on until the 26th letter (Z)=25
So F is the 6th letter, which would be index 5 in a zero-based array. So yes, F=5.

Similarly:
J is the 10th letter, so 9.
W is the 23rd letter, so 22.
Z is the 26th letter, so 25.

Therefore, I think that’s correct. There doesn’t seem to be any ambiguity unless there’s another rule at play, but with the information given, this seems like the logical conclusion.

F, J, W, and Z correspond to the numbers 5, 9, 22, and 25 respectively when each letter is assigned a number starting from A=0.

F = 5
J = 9
W = 22
Z = 25

Answer: F is 5, J is 9, W is 22, and Z is 25.

✅ deepseek-r1:32b

18 minutes and 38 seconds

Reply

Thoughts

Okay, so I’m trying to solve this problem where if A equals zero, I need to figure out the numbers corresponding to F, J, W, and Z. Hmm, not sure exactly how they’re connected yet. Maybe it’s related to some code or cipher?

Let me think… Sometimes letters can represent digits in a puzzle. If A is 0, maybe each letter stands for a unique digit from 0-9. But there are only four other letters: F, J, W, Z. So maybe they’re also single-digit numbers?

Wait, the problem just says “the number of F, J, W and Z.” That makes me think it’s about how many times each letter appears in something, like a word or an equation. But without more context, I’m not sure.

Maybe it’s a cryptarithmetic puzzle where each letter stands for a digit, and we have to solve for them based on some arithmetic. For example, A + B = C kind of thing. Since A is 0, perhaps substituting that in would help find the others.

Alternatively, could it be about something like the position in the alphabet? But F is 6, J is 10, W is 23, Z is 26. That seems too straightforward unless there’s a specific reason to think they’re related to A=0.

Wait, maybe it’s about Roman numerals or another numeral system where letters correspond to numbers differently. For example, in Roman numerals, I=1, V=5, etc., but F isn’t a standard Roman numeral symbol. Hmm, not sure that fits.

Another thought: sometimes in puzzles, letters represent digits with certain constraints. Since A is 0, maybe each of the other letters (F, J, W, Z) also represents unique digits from 1-9. But without an equation or more context, it’s hard to deduce their exact values.

Wait, could this be related to a specific word or phrase where A=0 and we have to find how many times F, J, W, Z appear? Maybe in the alphabet or something like that.

Alternatively, maybe each letter corresponds to its position in reverse. So if A is 1, but here it’s 0, perhaps subtracting from a certain number. That might complicate things though without more info.

I’m stuck because I don’t have enough information about how these letters are connected or what they represent numerically beyond A=0. Maybe there’s an equation involving F, J, W, Z that uses A somewhere? Without knowing the structure of that equation, it’s hard to solve for each variable.

Alternatively, perhaps it’s a substitution cipher where each letter is replaced by its position in the alphabet minus one since A=0. So B would be 1, C=2, etc. If that’s the case, then F would be 5 (since A=0), J would be 9, W=22, Z=25. That might make sense if it’s a positional cipher where each letter is shifted down by one.

Wait, in standard alphabet position, A=1, B=2,… but here A=0, so F=6-1=5, J=10-1=9, W=23-1=22, Z=26-1=25. That seems consistent if we’re adjusting for the zero-based index.

So maybe each letter’s numerical value is its position in the alphabet minus one. Therefore:

A = 1 – 1 = 0

F = 6 – 1 = 5

J = 10 – 1 = 9

W = 23 – 1 = 22

Z = 26 – 1 = 25

That could be a way to map them. But I’m not entirely sure if that’s what the problem is asking because it depends on the context of how these letters are being used.

Alternatively, maybe each letter corresponds to its Scrabble tile points or something like that? Let me recall: F is 4, J=8, W=4, Z=10. But A=1 in Scrabble, so if A=0 here, maybe it’s adjusted by subtracting 1 point from each letter’s value. So:

F = 4 – 1 = 3

J = 8 – 1 =7

W =4-1=3

Z=10-1=9

But that would mean F and W are both 3, which might not be what the problem wants since they probably expect unique numbers. So maybe not.

Another angle: sometimes in puzzles, letters can represent binary or hexadecimal values. For example, in hex, A is 10, but here it’s 0. Maybe each letter represents its hex value minus something? Not sure if that applies.

Wait, another idea: perhaps the number of segments in a seven-segment display for each digit. If A=0, which has six segments lit up (assuming standard digits). Then F is… Let me think: 0 uses six segments, so maybe A=6? But the problem says A=0. Hmm, confusing.

Alternatively, if A represents zero as in a count of something, maybe how many times each letter appears in a certain set or word. Without more context, it’s hard to say.

Wait, going back, perhaps it’s about alphabetical positions but with a twist. If A is 1 normally, here it’s 0, so the mapping could be off by one. So F would be position 6 – 1 =5, J=9, W=22, Z=25 as I thought earlier.

Alternatively, maybe each letter corresponds to the number of letters between them and A. For example, from A(1) to F is E, which is 5 letters apart, so F=5? Similarly, J is 9 letters after A. W is 23, Z is 26. So that also gives F=5, J=9, W=22, Z=25.

But without knowing the exact rule, it’s hard to confirm. Maybe I should consider standard mappings or common puzzles where letters are assigned numbers this way.

Another possibility: in ASCII codes, each letter has a numerical value. A is 65, B=66,… F=70, J=74, W=87, Z=90. But if A=0 here, maybe we subtract 65 from each, so:

F =70-65=5

J=74-65=9

W=87-65=22

Z=90-65=25

That gives the same results as before. So if A is mapped to 0 by subtracting its ASCII value from itself, then F would be 5, J=9, W=22, Z=25.

This seems plausible because it’s a common way to encode letters numerically in programming and puzzles. By assigning each letter a number relative to A being zero, you can map the entire alphabet accordingly.

So putting it all together:

If A = 0,

F = position of F – position of A +1? Wait no, since A=0, maybe just subtracting 1 from their standard positions.

Wait in ASCII, A is 65. If we set A=0, then each letter’s value would be its ASCII code minus 65. So:

F:70-65=5

J:74-65=9

W:87-65=22

Z:90-65=25

Yes, that makes sense.

Alternatively, in terms of zero-based index in the alphabet, A is 0, B=1,… Z=25. So F would be 5, J=9, W=22, Z=25.

That seems consistent with both methods: either subtracting 65 from ASCII or considering a zero-based index for letters.

Therefore, I think the numbers are as follows:

F =5

J=9

W=22

Z=25

Now, let’s break it down step by step:

Understand the given information : A is equal to 0.
Determine the mapping method :
- Since A corresponds to 0, we can use a zero-based index where each letter’s position in the alphabet minus one gives its numerical value.
Calculate the values for F, J, W, Z :
- F is the 6th letter (A=1), so 6-1=5.
- J is the 10th letter, so 10-1=9.
- W is the 23rd letter, so 23-1=22.
- Z is the 26th letter, so 26-1=25.

Answer : F = 5, J = 9, W = 22, Z = 25.

Numeric Comparison

In this test we will compare two numbers that are very similar. There are some LLMs that look at 0.45 and 0.5 and say that ’45’ is bigger. In this case though, every model of DeepSeek got it right first try.

Prompt

Let’s say I have two numbers, one being 305.45 and the other being 305.5, which one is bigger?

✅ deepseek-r1:1.5b

Took 4 seconds.

Reply

Thoughts

First, I need to compare the two numbers: 305.45 and 305.5.

Both numbers have the same whole number part, which is 305.

Next, I’ll examine the decimal parts:

For 305.45, the decimal portion is 0.45.
For 305.5, the decimal portion is 0.5.

Since 0.5 is greater than 0.45, 305.5 is larger than 305.45.

Solution:
To determine which number is larger between 305.45 and 305.5, let’s compare them step by step.

Compare the Whole Number Part:
- Both numbers have the whole number part as 305 .
Compare the Decimal Parts:
- For 305.45 : The decimal part is 0.45 .
- For 305.5 : The decimal part can be written as 0.5 , which is equivalent to 0.50 .
Determine Which Decimal is Larger:
- Comparing 0.45 and 0.50 , it’s clear that 0.50 is greater than 0.45 .

Since the decimal parts of both numbers show that 0.50 is larger, we can conclude that: $305.5$ $is$ $bigger$

✅ deepseek-r1:7b

Took 9 seconds.

Reply

Thoughts

First, I need to compare the two numbers: 305.45 and 305.5.

I’ll start by aligning their decimal places to make an accurate comparison.

Next, I’ll look at each corresponding digit from left to right.

Both numbers have the same integer part (305).

Then, comparing the tenths place: both have a 4 in this position.

Moving on to the hundredths place: one number has a 5 and the other has no second decimal place, which is equivalent to 0.

Since 5 is greater than 0, 305.45 is larger.

To determine which number is bigger between 305.45 and 305.5 , let’s compare them step by step.

Align the Decimal Places:
To make a fair comparison, it’s helpful to write both numbers with the same number of decimal places.
$305.45 305.50$
Compare Digit by Digit:
- Integer Part: Both numbers have the integer part 305 .
  - $305 = 305$
- Tenths Place: Compare the first digit after the decimal point.
  - $4$ (from 305.45) vs. $5$ (from 305.50)
  Since $4 < 5$ , we can already conclude that: $305.45 < 305.50$
- - $5$ (from 305.45) vs. $0$ (from 305.50)
    Hundredths Place: Even though this step isn’t necessary once the tenths place is determined, for completeness:
Conclusion:
Based on the comparison: $305.45 < 305.5$
Therefore, the bigger number is: 305.5

✅ deepseek-r1:8b

Took 9 seconds. I tried 10 times but it always got almost everything wrong, the reply below was the closest it got.

Reply

Thoughts

First, I compare the whole number parts of both numbers.

Both numbers have the same whole number part, which is 305.

Next, I examine the decimal parts.

The first number has .45, while the second number has .50.

Since 0.50 is greater than 0.45, the second number, 305.50, is larger.

Solution:
To determine which of the two numbers is bigger:

Compare the Whole Numbers:
- Both numbers have the same whole number part: 305.45 and 305.50
Examine the Decimal Parts:
- Compare the first decimal digits:
  - First number: 0.45
  - Second number: 0.50

- Since 0.50>0.45, the second number is larger.

Final Answer: 305.5 is bigger

✅ deepseek-r1:14b

Took 48 seconds.

Reply

Thoughts

First, I need to compare the two numbers: 305.45 and 305.5.

Both numbers have the same whole number part, which is 305.

Since the whole numbers are equal, I’ll look at the decimal parts. The first number has 45 hundredths, while the second number has 5 tenths, which is equivalent to 50 hundredths.

Comparing the decimal parts, 50 hundredths is greater than 45 hundredths.

Therefore, 305.5 is larger than 305.45.

To determine which number is bigger between 305.45 and 305.5 , follow these steps:

Compare the Whole Number Part:
- Both numbers have the same whole number part, which is 305 .
Compare the Decimal Parts:
- For 305.45 : The decimal part is 45 hundredths .
- For 305.5 : The decimal part is 5 tenths , which is equivalent to 50 hundredths .
Determine Which Decimal Part is Larger:
- 50 hundredths (from 305.5) is greater than 45 hundredths (from 305.45) .

Therefore, 305.5 is larger than 305.45 .

✅ deepseek-r1:32b

3 minutes and 50 seconds

Reply

Thoughts

First, I will compare the integer parts of both numbers. Both are 305, so they are equal in this aspect.

Next, I’ll look at the decimal parts: 0.45 for the first number and 0.5 for the second number. To make it easier to compare, I can express 0.5 as 0.50.

Now, comparing 0.45 and 0.50, it’s clear that 0.50 is larger because 50 hundredths is greater than 45 hundredths.

Therefore, the number 305.5 is bigger than 305.45.

Solution: Comparing $305.45$ and $305.5$

To determine which number is larger between $305.45$ and $305.5$ , follow these steps:

Compare the integer parts:
- Both numbers have the same integer part: $305$ .
Compare the decimal parts:
- For $305.45$ : The decimal part is $0.45$ .
- For $305.5$ : The decimal part is $0.5$ (which can be written as $0.50$ for easier comparison).
Determine which decimal is larger: 0.45<0.50

Since the decimal part of $305.5$ is greater than that of $305.45$ , the number $305.5$ is larger.

Final Answer: $305.5$

Strawberry Rs and Peppermint Ps

Now we will do a funny test where we give two words and ask it to tell us how many of a certain letter and vowels there are. As you will be able to see in the replies, most models got it wrong.

Prompt

In the word strawberry, how many of it’s letters are vowels and how many of them are Rs? Also, on the word peppermint how many letters are vowels and how many Ps are there?

❌ deepseek-r1:1.5b

Took 3 seconds on average. Tried 10 times, always gave atleast one of the numbers wrong.

Reply

Thoughts

First, I need to analyze the word “strawberry.” I will go through each letter one by one, checking whether it is a vowel or not. Additionally, I’ll identify if any of these letters are ‘R.’ Once I’ve done this for both words, I can provide the required information.

Strawberry: Vowels: 3, Rs: Not present

Peppermint: Vowels: 2, Ps: Yes

5️⃣ deepseek-r1:7b

Took on average 60 seconds. Tried 5 times, in one of the tries it thought so much about strawberries that it forgot about the peppermint, in another it thought so much about peppermint that it thought I also asked about ‘independent’ and on another try it thought so much (more than double the previous times) that it forgot to give an anwser. As you will see below, the model did make some mistakes, but it gave the correct numbers, so we will count that as a win.

Reply

Thoughts

Alright, let me try to figure out this problem. So I have two words: “strawberry” and “peppermint.” For each word, I need to determine how many vowels there are and how many times the letter ‘R’ appears in the first word and the letter ‘P’ appears in the second.

Starting with “strawberry.” First, let me write out all the letters so I can see them clearly: S, T, R, A, W, B, E, R, R, Y. Hmm, actually, wait, is that right? Let’s count: ‘s’, ‘t’, ‘r’, ‘a’, ‘w’, ‘b’, ‘e’, ‘r’, ‘r’, ‘y.’ So that’s 10 letters in total.

Now, vowels are the letters A, E, I, O, U. Sometimes Y is considered a vowel, but for now, let me just stick with the standard vowels: A, E, I, O, U. So looking at “strawberry,” which letters are vowels?

Starting from the beginning: S – consonant, T – consonant, R – consonant, A – vowel (count 1), W – consonant, B – consonant, E – vowel (count 2), R – consonant, R – consonant, Y – sometimes vowel. Hmm, so depending on whether I consider Y a vowel or not.

Wait, the problem didn’t specify if we include Y as a vowel. It just says vowels, which are typically A, E, I, O, U. So maybe I should treat Y as a consonant here? Or does it count?

Looking back at the question: “how many of its letters are vowels.” The word is “strawberry,” and in my mind, when we say vowels in words, usually only A, E, I, O, U are considered. So perhaps Y isn’t counted as a vowel here.

But wait, sometimes in riddles or specific contexts, Y can be considered a vowel. Hmm, but in standard English, it’s not typically included unless specified otherwise. So maybe Y is just treated as a consonant here. So in “strawberry,” the vowels are A and E. That would make two vowels.

Now, how many Rs are there? Let me go back through the word: S-T-R-A-W-B-E-R-R-Y. So I see R at position 3, then later another R at position 8, and another one at position 9. So that’s three Rs in “strawberry.” Wait a minute, so two vowels (A and E) and three Rs.

Wait, but hold on: when I count the letters again: S(1), T(2), R(3), A(4), W(5), B(6), E(7), R(8), R(9), Y(10). So yeah, three Rs. Okay, so that’s for “strawberry.”

Moving on to the second word: “peppermint.” Let me write down all the letters again: P, E, P, P, M, I, N, C, T, O, R, T? Wait, hold on. Hmm, maybe my initial breakdown is wrong.

Wait, let’s spell it out properly. P-E-P-P-M-I-N-C-T-O-R. So that should be 12 letters: P(1), E(2), P(3), P(4), M(5), I(6), N(7), C(8), T(9), O(10), R(11), T(12). Wait, but in the problem statement, it’s “peppermint.” Let me check: P-E-P-P-M-I-N-C-T-O-R. Yeah, that seems correct.

So letters are: 1:P, 2:E, 3:P, 4:P, 5:M,6:I,7:N,8:C,9:T,10:O,11:R,12:T

Okay, so vowels in “peppermint” would be E, I, O. Let’s count:

E at position 2 – vowel (count 1),

I at position 6 – vowel (count 2),

O at position 10 – vowel (count 3).

So that makes three vowels.

Now, how many Ps are in “peppermint”? Let me look through the letters again: P is at positions 1, 3, and 4. So that’s three Ps.

Therefore, in “strawberry,” there are two vowels and three Rs; in “peppermint,” there are three vowels and three Ps.

Wait a minute—hold on—is that correct? Let me verify once more for both words to make sure I didn’t make any mistakes.

Starting with “strawberry”:

Letters: S, T, R, A, W, B, E, R, R, Y

Vowels: A and E. So 2 vowels, correct.

Rs: The letters at positions 3,8,9 are Rs. That’s three Rs. Correct.

Now “peppermint”:

Letters: P, E, P, P, M, I, N, C, T, O, R, T

Vowels: E, I, O – three vowels.

Ps: The first letter is P (position 1), then positions 3 and 4. So that’s a total of three Ps? Wait—no! Because in the breakdown above, it’s positions 1,3,4—all Ps. That makes three Ps, right?

Wait, but hang on, let me count them: first letter is P (count 1), second E (not P), third P (count 2), fourth P (count 3). So yeah, the number of Ps is three.

So that seems correct.

But just to make sure I’m not making a mistake with “peppermint.” Let’s write out all letters:

1: P

2: E

3: P

4: P

5: M

6: I

7: N

8: C

9: T

10: O

11: R

12: T

So positions with Ps are 1,3,4. That’s three letters.

Alright, so that seems correct.

Therefore, the final counts:

“strawberry”: vowels = 2, Rs = 3
“peppermint”: vowels = 3, Ps = 3

So I think that’s it!

Final Answer In the word strawberry, there are \boxed{2} vowels and \boxed{3} Rs. In the word peppermint, there are \boxed{3} vowels and \boxed{3} Ps.

To determine how many vowels and specific letters are in each word, we analyze each letter systematically.

For the word “strawberry”:

The letters are: S, T, R, A, W, B, E, R, R, Y.
Vowels (A, E) are found at positions 4 and 7. Therefore, there are 2 vowels.
The letter ‘R’ appears at positions 3, 8, and 9. Therefore, there are 3 Rs.

For the word “peppermint”:

The letters are: P, E, P, P, M, I, N, C, T, O, R, T.
Vowels (E, I, O) are found at positions 2, 6, and 10. Therefore, there are 3 vowels.
The letter ‘P’ appears at positions 1, 3, and 4. Therefore, there are 3 Ps.

Final Answer:
In the word strawberry there are 2 vowels and 3 Rs and in the word peppermint there are 3 vowels and 3 Ps.

❌ deepseek-r1:8b

Took 9 seconds. I tried 10 times but it always got almost everything wrong, the reply below was the closest it got.

Reply

Thoughts

First, I will examine each word individually.

For “strawberry”:

Vowels : The letters are S, T, R, A, W, B, E, R, R, Y.
Identifying Vowels : Among these, the vowels are A and E. So there are 2 vowels.
Identifying Rs : There are three R’s in the word.

For “peppermint”:

Vowels : The letters are P, E, P, P, M, I, N, T.
Identifying Vowels : The vowels are E and I. So there are 2 vowels.
Identifying Ps : There are three P’s in the word.

Let’s analyze each word step by step.

1. Word: Strawberry

Letters: S, T, R, A, W, B, E, R, R, Y
Vowels: Identify the vowels in the word.
- A and E are vowels.
- Total Vowels: 2
Number of Rs:
- Count the letters R.
- There are three Rs.
- Number of Rs: 3

2. Word: Peppermint

Letters: P, E, P, P, M, I, N, T
Vowels: Identify the vowels in the word.
- E and I are vowels.
- Total Vowels: 2
Number of Ps:
- Count the letters P.
- There are three Ps.
- Number of Ps: 3

Final Answer

For Strawberry :

Vowels: $2$
Rs: $3$

For Peppermint :

Vowels: $2$
Ps: $3$

❌ deepseek-r1:14b

Took 17 seconds. Tried 5 times.

Reply

Thoughts

First, I’ll examine the word “strawberry.” It has 9 letters. The vowels in this word are ‘a’ and ‘e,’ so there are 2 vowels.

Next, I’ll look for the letter ‘R.’ There’s one ‘R’ in “strawberry.”

Moving on to the word “peppermint,” it consists of 10 letters. The vowels here are ‘e,’ ‘i,’ and two ‘e’s, making a total of 4 vowels.

Lastly, I’ll count how many times the letter ‘P’ appears in “peppermint.” There are two ‘P’s at the beginning of the word.

1. The Word “strawberry”

Total Letters: 9
Vowels Identification:
- Vowels in English are A, E, I, O, U (both uppercase and lowercase).
- In the word s t r a w b e r r y :
  - a , e are vowels.
Count of Vowels: 2
Count of ‘R’:
- The letter r appears twice in “strawberry.”

2. The Word “peppermint”

Total Letters: 10
Vowels Identification:
- In the word p e p p m i n t :
  - e , i , and two e’s are vowels.
Count of Vowels: 4
Count of ‘P’:
- The letter p appears twice at the beginning of “peppermint.”

Final Answer:

In “strawberry”:
- Vowels = 2
- Rs = 2
In “perppermint”:
- Vowels = 4
- Ps = 2

❌ deepseek-r1:32b

Took on average 5 minutes and 29 seconds. Tried 10 times.

Reply

Thoughts

First, I will examine the word “strawberry.”

I’ll list each letter to identify the vowels. The vowels in English are a, e, i, o, u.

Letters in “strawberry”: s, t, r, a, w, b, e, r, y

Vowels: a, e

So, there are 2 vowels.

Next, I’ll count the number of ‘R’s in “strawberry.”

Looking at the letters again: s, t, r, a, w, b, e, r, y

There are two ‘r’s present.

Now, moving on to the word “peppermint.”

I’ll list each letter and identify the vowels.

Letters in “peppermint”: p, e, p, p, m, i, n, t

Vowels: e, i

This gives us 2 vowels.

Finally, I’ll count the number of ‘P’s in “peppermint.”

From the letters: p, e, p, p, m, i, n, t

There are three ‘p’s present.

1. The Word “Strawberry”

Letters: S, T, R, A, W, B, E, R, Y
Vowels in English: A, E, I, O, U

Counting Vowels:

A (position 4)
E (position 7)

Total Vowels: 2

Counting ‘R’s:

R (position 3)
R (position 8)

Total ‘R’s: 2

2. The Word “Peppermint”

Letters: P, E, P, P, M, I, N, T
Vowels in English: A, E, I, O, U

Counting Vowels:

E (position 2)
I (position 6)

Total Vowels: 2

Counting ‘P’s:

P (position 1)
P (position 3)
P (position 4)

Total ‘P’s: 3

Final Answer:

Strawberry: Vowels: 2, Rs: 2
Peppermint: Vowels: 2, Ps: 3

Positional and time awareness

In this test we will give DeepSeek a schedule and ask where the cat is at a certain time. The point of this test is to get a reply where DeepSeek tells us where that cat is and what it is doing, not only ‘what’.

Prompt

Every day from 2PM to 4PM the household cat, Tobias, is in the window. From 2 until 3, Tobias is looking at birds. For the next half hour, Tobias is sleeping. On the final half hour, Tobias is cleaning himself. The time is 3:14PM, where and what is Tobias doing.

9️⃣ deepseek-r1:1.5b

Took 4 seconds on average. Tried 9 times. Most of the tries DeepSeek told me that the cat was looking at birds while cleaning himself, which is incorrect.

Reply

Thoughts

First, I need to understand the timeline of Tobias’ activities during the day when he’s in the window from 2PM to 4PM.

During this period, between 2PM and 3PM, Tobias is observing birds through his window.

From 3:00PM to 3:30PM, he is asleep. This makes sense as it’s a half-hour time slot.

After 3:30PM until 4:00PM, the final half-hour of his time in the window is dedicated to cleaning himself.

Given that the current time is 3:14PM, I can determine which activity Tobias was doing by comparing this time with his allocated activities.

Since 3:00PM to 3:30PM falls within the sleeping period, at 3:14PM, Tobias must be asleep.

At 3:14 PM , Tobias is asleep in his window.

2️⃣ deepseek-r1:7b

Took 5 seconds on average. On the first try it only told me that the cat was sleeping, at the second time it got the correct anwser.

Reply

Thoughts

Okay, so I need to figure out where and what the cat Tobias is doing at 3:14 PM based on the given schedule.

First, let me parse the information:

From 2 PM to 4 PM every day, the household cat, Tobias, is in the window.

The activities are broken down as follows:

From 2 PM until 3 PM (which is one hour), he’s looking at birds. So that’s from 2:00 PM to 3:00 PM.

Next half hour—so starting right after 3 PM, for the next 30 minutes, he’s sleeping. That would take him from 3:00 PM to 3:30 PM.

Then, on the final half hour before 4 PM, which is from 3:30 PM to 4:00 PM, he’s cleaning himself.

Now, the current time given is 3:14 PM. I need to see where this falls within the schedule.

So breaking it down:

The first hour (2-3 PM): birdwatching.

At exactly 3 PM, that period ends. Then comes the sleep time from 3 PM to 3:30 PM. Since 3:14 PM is after 3 PM but before 3:30 PM, it should be during his sleep.

Wait a minute—does “the next half hour” start exactly at 3 PM? The way it’s written says “from 2 until 3, then for the next half hour,” which would imply that the birdwatching is from 2-3 PM, and immediately after, starting at 3 PM, he sleeps for the next 30 minutes.

Yes, so 3:14 PM is within the sleep period. Therefore, he’s sleeping.

So the answer should be that at 3:14 PM, Tobias is in the window, asleep.

At 3:14 PM, the cat Tobias is in the window, sleeping.

2️⃣ deepseek-r1:8b

Took 9 seconds on average. Same as with the 7b model. First try only told me what the cat was doing, second try nailed it.

Reply

Thoughts

Okay, so I need to figure out where and what Tobias, the household cat, is doing at 3:14 PM. Let me break this down step by step.

First, let’s outline Tobias’ daily schedule based on the information provided:

From 2 PM to 3 PM : Tobias is looking at birds.
From 3 PM to 3:30 PM : Tobias is sleeping.
From 3:30 PM to 4 PM : Tobias is cleaning himself.

The time in question is 3:14 PM, which falls within the second and third intervals because it’s after 3 PM but before 3:30 PM.

Wait, actually, let me check that again. The timeline given is from 2 PM to 4 PM, divided into three parts:

2 PM to 3 PM: Looking at birds.
3 PM to 3:30 PM: Sleeping.
3:30 PM to 4 PM: Cleaning.

So, 3:14 PM falls between 3 PM and 3:30 PM. That means Tobias is in the second part of his schedule, where he’s sleeping. But wait, the third interval starts at 3:30 PM, so before that, from 3 PM to 3:30 PM, he’s asleep.

Wait a second, let me make sure I’m interpreting this correctly. The first half-hour is 2-3 PM, looking at birds. Then the next half hour is 3-3:30 PM, sleeping. And the final half hour is 3:30-4 PM, cleaning. So yes, at 3:14 PM, which is within 3 PM to 3:30 PM, Tobias is asleep.

So, putting it all together, the time is 3:14 PM, and based on his schedule, he’s in the window, asleep.

Answer: At 3:14 PM, Tobias is asleep in the window.

The cat’s schedule is as follows:

2 PM to 3 PM: Looking at birds.
3 PM to 3:30 PM: Sleeping.
3:30 PM to 4 PM: Cleaning himself.

✅ deepseek-r1:14b

Took 26 seconds. Nailed it first try.

Reply

Thoughts

Okay, so I need to figure out what Tobias the cat is doing at 3:14 PM based on the given schedule.

First, let me break down the timeline from 2 PM to 4 PM.

From 2 PM to 3 PM, that’s one hour. During this time, Tobias is looking at birds. So, for the first half of his window, he’s bird-watching.

Then, after 3 PM, there are two more hours until 4 PM. The user mentioned that for the next half hour, he’s sleeping, and then on the final half hour, he’s cleaning himself. Wait, but that adds up to an additional hour (half-hour sleep + half-hour cleaning). So from 3 PM onwards, let me structure this.

From 3 PM to 3:30 PM, Tobias is sleeping. That’s the next half hour after bird-watching. Then, from 3:30 PM to 4 PM, he’s cleaning himself for the final half hour.

So putting it all together:

2 PM – 3 PM: Looking at birds.
3 PM – 3:30 PM: Sleeping.
3:30 PM – 4 PM: Cleaning himself.

Now, the time in question is 3:14 PM. Let’s see where that falls. It’s after 3 PM but before 3:30 PM. Specifically, 3:14 PM is within the sleeping period because it hasn’t reached 3:30 PM yet.

So at 3:14 PM, Tobias should be sleeping.

At 3:14 PM, Tobias is sleeping in the window during his scheduled naptime from 3 PM to 3:30 PM.

✅ deepseek-r1:32b

Took 5 minutes and 23 seconds. Like the 14b model, gave the correct anwser the first time.

Reply

Thoughts

Okay, so I need to figure out what Tobias, the household cat, is doing at 3:14 PM. Let me break this down step by step.

First, I know that every day from 2 PM to 4 PM, Tobias is in the window. So his activities are confined within this two-hour window. The problem divides this time into three segments:

From 2 PM until 3 PM: Looking at birds.
The next half hour (which would be from 3 PM to 3:30 PM): Sleeping.
The final half hour (from 3:30 PM to 4 PM): Cleaning himself.

Now, the question is asking about the time at 3:14 PM. Let me think about where this falls within the schedule.

From 2 PM to 3 PM, that’s one hour. So up until 3 PM, he’s looking at birds. Then, from 3 PM onwards, he starts sleeping for the next half hour, which is until 3:30 PM. After that, from 3:30 PM to 4 PM, he cleans himself.

So, 3:14 PM is after 3 PM but before 3:30 PM. That means it’s within the second segment of his schedule when he’s sleeping.

Wait a minute, let me double-check the timings to make sure I didn’t mix up any numbers. The first activity is from 2-3 PM, which is clear. Then next half hour would be 3 PM to 3:30 PM for sleeping, and then cleaning himself from 3:30 PM to 4 PM.

Yes, that adds up correctly: 1 hour (looking at birds) + 0.5 hours (sleeping) + 0.5 hours (cleaning) equals exactly two hours from 2 PM to 4 PM.

So at 3:14 PM, which is 14 minutes past 3 PM, Tobias would be in the middle of his sleeping period because that’s when he sleeps until 3:30 PM.

I don’t think there’s any overlap or confusion here. The timeline is straightforward:

Before 3 PM: Looking at birds.
From 3 PM to 3:30 PM: Sleeping.
From 3:30 PM onwards: Cleaning himself.

Therefore, 3:14 PM falls clearly in the sleeping period.

At 3:14 PM, Tobias is in the window and is currently sleeping. This falls within his scheduled sleep time from 3 PM to 3:30 PM.

Answer: At 3:14 PM, Tobias is sleeping in the window.

100 Digits of π

Some models gave me the digit 3 + 100 decimal digits others gave me the digit 3 + 99 decimal digits, I counted both as correct.

Prompt

Tell me the first 100 digits of pi.

✅ deepseek-r1:1.5b

Took 8 seconds.

Reply

Thoughts

Okay, so I need to figure out how to find the first 100 digits of pi. Hmm, I know that pi is an irrational number, which means it goes on forever without repeating. But just knowing that doesn’t help me get specific digits. Let’s break this down.

First, I remember that pi starts with 3.14159265358979323846… and so on. That’s the first few digits. So, to get the next ones, maybe there are methods or formulas for calculating pi to a large number of decimal places. I’ve heard about something called Machin’s formula before. Let me think if that’s one.

Machin’s formula is used to compute pi by using an arctangent series. It goes like this: pi = 4*(arctan(1/5) – arctan(1/239)). To calculate these arctangents, I believe there are Taylor series expansions that can be used. The Taylor series for arctan(x) is x – x^3/3 + x^5/5 – x^7/7 + … and so on.

So, if I plug in 1/5 and 1/239 into this series, I should get the values of arctan(1/5) and arctan(1/239). Then subtract them, multiply by 4, and that should give me pi. But wait, how many terms do I need in each series to get enough precision?

I think for high precision, you might need to use more terms in the Taylor series. Maybe around 50 or 70 terms? That seems like a lot, but it’s manageable computationally. Since I’m doing this manually here, perhaps there’s another approach.

Wait, maybe I can find a known value of pi up to 100 digits online. But since I don’t have access right now, that might not be feasible. Alternatively, if I had some computational tools or access, I could look into algorithms like the Gauss-Legendre algorithm or Chudnovsky’s formula for more efficient calculations.

The Chudnovsky algorithm is even better because it converges extremely quickly. It involves factorial and prime number terms, but since I don’t have a calculator with those functions, that might not be practical either. However, maybe there are approximations or known sequences that can help me get the digits without calculating from scratch.

Another thought: sometimes mathematical constants have their decimal expansions stored as strings in databases or programming libraries. For example, the first 100 digits of pi are commonly found online with precise accuracy. But again, I don’t know if I have access to such resources right now.

Alternatively, perhaps I can use an approximation formula that truncates at a sufficient number of terms. Let me consider using the Leibniz formula for pi: pi = 4*(1 – 1/3 + 1/5 – 1/7 + …). But again, this converges very slowly. For 100 digits, I would need to calculate about 25,000 terms or something like that.

Wait a minute, maybe I can find the first few digits and then use some patterns or additional formulas for the rest. I know pi is transcendental, so its decimal expansion is non-repeating and non-terminating. That makes it hard to predict further digits without computing them directly.

Let me also consider if there’s any known value of pi in hexadecimal or other bases that might help. For example, in hexadecimal, pi is 3.243F6A8079… but converting this to decimal would complicate things unless I know the exact sequence up to 100 digits.

Another idea: perhaps using a formula known for generating pi’s decimal expansion digit by digit, like the Brent-Salamin algorithm or something similar. But implementing that without a calculator is challenging manually.

Wait, maybe if I recall any memorized digits of pi and then look up the rest? I think some people memorize them, especially with certain tricks to extend the number beyond known points. For example, using the Major System or other memory techniques. However, trying to do that for 100 digits manually would take a lot of time and requires more advanced knowledge.

Alternatively, if I have access to a computer algebra system or a calculator, I could input the formula directly into it to get pi up to 100 decimal places. But since I’m thinking about this as a student, maybe relying on known algorithms is better for learning purposes.

Let me also think about how computers compute pi. They use binary representations and fast Fourier transforms, but again, that’s beyond my manual calculation capabilities. Still, knowing the process might help me understand why there are so many digits after pi starts with 3.14…

So, putting this all together: without an advanced computational tool, it’s very tedious to calculate pi up to 100 digits manually. I can at least remember that it starts with 3.14159265358979323846… and then use additional methods or algorithms for the rest, but each method has its own set of steps and may require significant computation.

Wait, maybe there’s a way to calculate pi using fractions or other approximations that have known decimal expansions. For instance, 355/113 is a common approximation, which gives about 6 or 7 correct digits, but not enough for 100. So perhaps I need higher-order approximations.

I think it’s also useful to note that the first few digits are well-known and can be memorized by some people, while the rest requires more work. There might even be some sources online listing pi up to 100 decimal places, but without access, that’s not helpful in this scenario.

In summary, manually calculating the first 100 digits of pi is challenging due to its non-repeating, non-terminating nature and the complexity of efficient algorithms needed. However, using known formulas like Machin’s or the Brent-Salamin algorithm allows for accurate computation once one has a computational tool.

The first 100 digits of pi are as follows:

3.1415926535897932384626433832795028841971693993751058209749445923078164062862089986280348253421170679

✅ deepseek-r1:7b

Took 2 seconds.

Reply

Thoughts

I need to provide the first 100 digits of π.

First, I recall that π is an irrational number with a non-repeating decimal sequence.

I remember most of its initial digits: 3.141592653589793…

To reach 100 digits, I’ll need to find the subsequent numbers after this point.

Using my knowledge and memorization, I can list out the digits step by step until I have exactly 100 digits.

Sure! Here are the first 100 digits of π (pi):
3.141592653589793238462643383279502884197169399375105820974944592307816406286208998628034825342117067

Let me know if you’d like further assistance! 😊

❌ deepseek-r1:8b

Took 5 minutes on average. DeepSeek tried to calculate all 100 digits one by one using a formula, it took a while… and it also gave the wrong anwser every time (10 tries). I did not attatch DeepSeek’s thoughts this time because of how big they were.

Reply

The first 100 digits of pi (π) are as follows: 3.14159265358979323846264338327950288419716939937510…
These digits have been well-established through extensive computational efforts and are commonly referenced in mathematical literature.

2️⃣ deepseek-r1:14b

Took on average 2 minutes and 52 seconds. Like the 8b model, it calculated the digits one by one, so like before I won’t append it’s thoughts. Got the correct reply in the second try, on the first one it only gave me 50 digits.

Reply

The decimal expansion of π (pi) up to its first 100 digits is as follows:

3.1415926535 8979323846 2643383279 5028841971 6939937510
5820974944 5923078164 0628620899 8628034825 3421170679

Breaking it down by positions:

1
4
1
5
9
2
6
5
3
5
… # DeepSeek lists every number until the 100th

❌ deepseek-r1:32b

Took almost 30 minutes and stopped thinking. Tried 3 times, same result. It got the number correctly in it’s thoughts, but because it doubted itself, I never got a reply, so I can’t mark it as a success.

Generating a smile SVG

In this test we are trying to get DeepSeek to give us the SVG code to something that looks like a smile emoji 🙂. For this question i didn’t test the 32b model because it was getting stuck in it’s own thoughts for a looong time. There was also some others that the thoughts were so long that i could not attach them below. As a comparison, you can see a smile generated by ChatGPT here:

Prompt

Generate the SVG code to draw a smile.

❌ deepseek-r1:1.5b

Took 6 seconds on average. Tried 10 times, none look anything like a smile. Most of them aren’t even valid SVG.

Reply

Thoughts

Alright, the user wants me to generate SVG code to draw a smile. I need to figure out how to approach this.

First, I should consider what elements make up a smile in an animation. Typically, it involves a mouth opening and closing with a smiley face. So, I’ll need to include features like a circular mouth, a rounded smile curve, and some eye parts.

I remember that SVG uses paths and shapes for such animations. Using the cubic Bezier curve could help create smooth animations because they can represent rotations easily. So, using two Bézier curves—like a mouth and an eye part—I think would work well.

Next, I’ll need to define the outline of the face. A circular shape seems appropriate. To make it look round, I might use circles with different radii for eyes. Maybe a smaller inner circle for the left eye and another slightly larger one for the right eye.

For the animation, I want the smile to move smoothly. Using two Bezier curves allows me to control their rotation. The outer curve can be rotated once clockwise, creating the opening of the mouth when animated. The inner curve should rotate twice to simulate a closed face opening into a smile as it moves forward.

I’ll structure this by defining the overall path for each curve and then adding stroke attributes with animation. This way, each part can rotate progressively over time, making the animation smooth.

Finally, I’ll put all these elements together in an SVG element and apply appropriate stroke colors to make it look like a smiley face. I should test the code to ensure it works correctly and looks good when animated.

Here’s an SVG code that creates a simple smile animation:

<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24">
  <path stroke="white" stroke-width="2" d="M16.5 0 C17.5 0 18 0.35 17 0.69c-1.33-.47-3.63-2.34-3.52-2.34-2.49-1.85-1.85-1.85-1.33-.47.47 1.4 2.47 2.47 2.53 3.02 2.53 3.02 2.47 2.47c2.47.47 3.52.89 3.63.45c1.32-.45 17-1 17-1.69c0-1.32 0.35-2.34 0.69-2.49L8.5 12l4-4"mxl="16" my="12"/>
  <path stroke="white" stroke-width="2" d="M14.5 7.55c-.13-.13-1.74-.38-3-1.01C17.9 6.01 18 6 17 6.99c.34.14.76.38 1.01.33L16 20l-2-2zm24 0v2h3v-2h3v-2z"/>
  <path stroke="white" stroke-width="2" d="M12 6C12 6.58 11.79 7.45 11 8.989zm10.54 6.973a2.927 2.927 0 0 2.035 0 5.703 2.927 5.703 0 0 0 0Z"/>
</svg>

❌ deepseek-r1:7b

Took 20 seconds on average. This was the closest thing to a smile that this model generated, the other options were all random lines:

Reply

Thoughts

Okay, I need to figure out how to generate SVG code for a smiley face. Let me break it down step by step.

First, I know that an SVG (Scalable Vector Graphics) file starts with the <svg> tag. I should set the width and height, maybe 200×200 pixels as a good size. Also, include a viewBox and any necessary attributes like stroke or fill properties.

Next, the main shape of a smiley face is usually a circle for the head and an arc for the eyes. So, I’ll create a circle with the <circle> element. The circle should be filled with a color, perhaps white or another color depending on the design.

Then, the eyes can be created using two small circles within the head. They might be filled with brown to look like real eyes. Positioning them correctly is important—probably symmetrically on either side of the center.

The smile itself comes from an arc that curves upwards. Using the <path> element allows for more control over the curve. I can define a Bezier curve here, specifying the control points to make it smiley. The path data (d attribute) will have the necessary values to create the smile shape.

I also want to add features like a small mouth and perhaps an ear. The mouth can be another arc within the head, maybe filled with a darker color for depth. An ear could be a simple shape, perhaps a small circle or rectangle attached to the side of the head.

Including details like wings is optional but adds more realism. They can be thin lines above the head, placed symmetrically on both sides.

Finally, I’ll make sure all elements are properly nested and that the SVG is correctly structured. Testing the code by saving it as an .svg file would help verify if everything displays as intended.

I should also consider using appropriate colors for each part to ensure the smiley face looks cohesive and recognizable.

Here’s the SVG code to draw a smile:

<?xml version="1.0" encoding="UTF-8"?>
<svg width="200" height="200" viewBox="0 0 200 200" xmlns="http://www.w3.org/2000/svg">
  <circle cx="100" cy="100" r="90" fill="white"/>
  
  <!-- Left eye -->
  <circle cx="80" cy="115" r="6" fill="brown"/>
  
  <!-- Right eye -->
  <circle cx="120" cy="115" r="6" fill="brown"/>
  
  <!-- Smile (made with path) -->
  <path d="M140,80 C179.15,80,203.5,94.49,208.44,108.43C210.62,121.72,198.9,135.25,176.95,140.27C155.68,142.33,131.62,135.72,108.93,128.43C92.48,123.81,85.77,115.93,80,100C73.27,88.69,85.66,77.93,102.67,65.46C116.56,66.96,136.19,81.71,140,80" fill="none" stroke="black"/>
  
  <!-- Mouth -->
  <path d="M100,82 A20,20 0 0,1 80,102 C75.66,93.34 70,85 75.66,76.66C81.34,68.34 85,77.5 80,87A20,20 0 0,1 60,102C54.34,93.34 59,85 54.34,76.66C58.66,68.34 65,77.5 70,87A20,20 0 0,1 80,107C81.34,93.34 85,85.5 80,80S80,77.66 78.66,76.34C77.34,75 75,72.5 77.34,70A12.59,12.59 0 0,1 78.1,62.22L80,60A12.59,12.59 0 0,1 78.1,57.78C77.34,56 75,52.5 77.34,50A12.59,12.59 0 0,1 80,60" fill="black"/>
  
  <!-- Left ear -->
  <circle cx="70" cy="100" r="5" fill="brown"/>
  
  <!-- Right ear (optional) -->
  <circle cx="130" cy="100" r="5" fill="brown"/>
  
  <!-- Wings (optional) -->
  <path d="M60,90 L40,120 L80,120" stroke="black" stroke-width="2"/>
  <path d="M140,90 L160,120 L120,120" stroke="black" stroke-width="2"/>
</svg>

❌ deepseek-r1:8b

Took 40 seconds on average. Still nothing usable.

Reply

Here’s a simple SVG representation of a smile:

<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 100 100" width="400" height="400">
  <title>Smile</title>
  
  <!-- Main circle -->
  <circle cx="50" cy="50" r="40" fill="white"/>
  
  <!-- Smile arc -->
  <arc cx="50" cy="50" r="30" startAngle="-90" endAngle="-30" fill="black"/>
  
  <!-- Teeth -->
  <g transform="rotate(0)">
    <line x1="20" y1="35" x2="40" y2="45" stroke="black" stroke-width="5"/>
    <line x1="50" y1="25" x2="60" y2="35" stroke="black" stroke-width="5"/>
    <line x1="80" y1="35" x2="90" y2="45" stroke="black" stroke-width="5"/>
    <line x1="30" y1="40" x2="50" y2="60" stroke="black" stroke-width="5"/>
    <line x1="70" y1="40" x2="100" y2="60" stroke="black" stroke-width="5"/>
  </g>
  
  <!-- Upper arc -->
  <path d="M 50,30 L 40,35 Q 20,25 0,15 A 10,10 0 0,1 60,15 Q 80,25 100,35 L 100,70" stroke="black" fill="none"/>
</svg>

❌ deepseek-r1:14b

Took 1 minute and 30 seconds. Finally we got something that looks like a smile, but sad. This was the only model that gave us something close enough to a smile, but it still ins’t one, so we can’t give it a ✅.

Reply

Here’s the SVG code to draw a smile:

<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 200 200">
  <!-- Face -->
  <circle cx="100" cy="100" r="80" fill="white" stroke="black" stroke-width="2"/>
  
  <!-- Left eye -->
  <circle cx="70" cy="85" r="12" fill="white" stroke="black" stroke-width="2"/>
  
  <!-- Right eye -->
  <circle cx="130" cy="85" r="12" fill="white" stroke="black" stroke-width="2"/>
  
  <!-- Mouth -->
  <path d="M90,120 Q100,120 110,130" 
         fill="none" 
         stroke="black" 
         stroke-width="4"/>
</svg>

Results

Now lets calculate a score for each model. Each ✅ gives one point, each 2️⃣ gives 1 / X points and ❌ do not give points.

Question	1.5b	7b	8b	14b	32b
Sentence parsing and counting	❌	✅	❌	✅	✅
Deductive arrays and offsets	2️⃣	✅	❌	✅	✅
Numeric Comparison	✅	✅	✅	✅	✅
Strawberry Rs and Peppermint Ps	❌	5️⃣	❌	❌	❌
Positional and time awareness	9️⃣	2️⃣	2️⃣	✅	✅
100 Digits of π	✅	✅	❌	2️⃣	❌
Generating SVGs	❌	❌	❌	❌	❌
Points	2.6	4.7	1.5	4.5	4

As you can see, in my tests, the 7b models was the best in terms of speed and correctness followed by the 14b model, which is still pretty fast and got a similar result to 7b, the 32b model sadly was too much for my computer and kept failing or just took too much time to reply, the 1.5b model was about what I expected but where I was a bit shocked was with 8b, I don’t know why, but this model was always losing itself in thought or just making stupid replies. Another thing I was a little disapointed on was with the SVG generation.

And that’s it. Thanks for reading and stay tuned for more tech insights and tutorials. Until next time, keep exploring the world of tech!