Updated Jul 23, 2025

Our Verdict

Mistral Devstral Medium

Qwen3 Coder

Too close to call. Pick based on your use case.

Pick Mistral Devstral Medium for Reasoning. Pick Qwen3 Coder for Web Design, Image Generation.

Too close to call

Writing DNA

Style Comparison

Similarity

92%

Qwen3 Coder uses 11.6x more emoji

Mistral Devstral Medium

Qwen3 Coder

50%Vocabulary61%

17wSentence Length72w

0.56Hedging0.61

4.5Bold4.7

5.3Lists5.6

0.00Emoji0.12

0.19Headings1.43

0.32Transitions0.20

Based on 21 + 22 text responses

vs

Ask them anything yourself

Mistral Devstral Medium

Qwen3 Coder

Some models write identically. You are paying for the brand.

178 models fingerprinted across 32 writing dimensions. Free research.

Model Similarity Index

185x

price gap between models that write identically

178

models

12

clone pairs

32

dimensions

Devstral M / S

95.7%

Qwen3 Coder / Flash

95.6%

GPT-5.4 / Mini

93.3%

Read the full report or download the 14-slide PDF

279 AI models invented the same fake scientist.

We read every word. 250 models. 2.14 million words. This is what we found.

AI Hallucination Index 2026

Free preview13 of 58 slides

Download the free preview or get all 58 slides for $49

FAQ

Common questions

Keep going

Mistral Devstral Medium vs GPT OSS 120BNew provider

Mistral Devstral Medium vs Pony AlphaNew provider

Mistral Devstral Medium vs MiniMax M2.5New provider

Mistral Devstral MediumvsQwen3 Coder

37 fights queued

Why Mistral Devstral Medium?

Dead even. This one's a coin flip.

Why Qwen3 Coder?

2.1x cheaper overall$0.22/M in · $0.95/M out

Leads 1 of 1 benchmarks

Stronger on SWE-bench Verified69.6% vs 61.6%

Mistral Devstral MediumQwen3 Coder

Input price

$0.40/M

$0.22/M

Output price

$2.00/M

$0.95/M

Released

Jul 2025

Jul 2025

Benchmarks1 common

SWE-bench Verified

61.6%

69.6%+11.5%

Favorites

Mistral Devstral MediumQwen3 Coder

Movie

The Shawshank Redemption

The Shawshank Redemption

The Shawshank Redemption

The Shawshank Redemption

Album

Kind of Blue

Kind of Blue

Book

The Name of the Wind

The Name of the Wind

Patrick Rothfuss

The Left Hand of Darkness

The Left Hand of Darkness

Ursula K. Le Guin

City

Tokyo

Kyoto

Game

The Legend of Zelda: Breath of the Wild

The Legend of Zelda: Breath of the Wild

Adventure, Action4.5

The Stanley Parable

The Stanley Parable

Indie, Adventure4.4

12 / 37

01

Estimate Complexityreasoning

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

I'll estimate the FLOPs required to train GPT-3 by breaking down the key components and making reasonable approximations.

Key Parameters of GPT-3

Parameters: ~175 billion (1.75 × 10¹¹)
Training tokens: ~300 billion (3 × 10¹¹)
Context window: 2048 tokens

Step 1: FLOPs per Token Forward Pass

For a transformer model, the forward pass FLOPs are dominated by matrix multiplications in the attention and MLP layers.

Attention layers:

Self-attention: ~4 × d_model × d_model × sequence_length per layer
With d_model ≈ 12,288 (based on GPT-3 architecture) and 96 layers
Attention FLOPs per token ≈ 96 × 4 × 12,288² ≈ 5.8 × 10¹⁰

MLP layers:

Each MLP has ~8 × d_model² FLOPs per token (two linear transformations)
MLP FLOPs per token ≈ 96 × 8 × 12,288² ≈ 1.16 × 10¹¹

Total forward pass FLOPs per token: ~1.7 × 10¹¹

Step 2: FLOPs per Token Backward Pass

Backward pass typically requires ~2× the forward pass FLOPs due to gradient computations.

Backward pass FLOPs per token: ~3.4 × 10¹¹

Step 3: Total FLOPs per Token

Forward + Backward = ~5.1 × 10¹¹ FLOPs per token

Step 4: Training FLOPs

However, we need to account for multiple epochs and the fact that training typically processes sequences, not individual tokens.

Sequence-based calculation:

Average sequence length: ~1024 tokens (conservative estimate)
Sequences processed: 3 × 10¹¹ tokens ÷ 1024 ≈ 2.9 × 10⁸ sequences
FLOPs per sequence: ~5.1 × 10¹¹ × 1024 ≈ 5.2 × 10¹⁴
Total training FLOPs: 2.9 × 10⁸ × 5.2 × 10¹⁴ ≈ 1.5 × 10²³

Step 5: Additional Considerations

Parameter updates and optimizer computations (Adam): ~2× the backward pass FLOPs
Multiple epochs (effective training): roughly 2-3× the base calculation

Final Estimate

Total training FLOPs: ~3 × 10²³ to 5 × 10²³ FLOPs

This is equivalent to approximately 300 to 500 exaFLOPs.

This estimate aligns reasonably with published figures and captures the massive computational requirements of training such a large language model, representing one of the most computationally intensive AI training runs to date.

Try this prompt

Vote

2 / 37

02

AI Board Game Logicreasoning

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025svg

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025svg

Try this prompt

Vote

3 / 37

03

Logic Puzzlereasoning

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

If Sally has 3 brothers and each brother has 2 sisters, it means that Sally is one of the sisters. Therefore, Sally has 1 sister.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

I need to work through this step-by-step.

Given information:

Sally has 3 brothers
Each brother has 2 sisters

Let me think about this carefully:

Since Sally is a girl, she is one of the sisters that each brother has.

If each brother has 2 sisters, and Sally is one of those sisters, then there must be 1 additional sister besides Sally.

Therefore, Sally has 1 sister.

To verify:

Sally + 1 sister = 2 sisters total
Each of the 3 brothers has these same 2 sisters
This matches the given information that each brother has 2 sisters

Sally has 1 sister.

Try this prompt

Vote

4 / 37

04

Math Misconception Testreasoning

Tests an AI's understanding of number representation

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

No, 9.11 is not greater than 9.9.

To compare these numbers, I can look at them digit by digit:

Both numbers have 9 in the units place
In the tenths place: 9.11 has 1, while 9.9 has 9
Since 1 < 9, we know that 9.11 < 9.9

Another way to see this is to write 9.9 as 9.90:

9.11 vs 9.90
The tenths digit (1 vs 9) determines the comparison

Therefore, 9.9 is greater than 9.11.

Try this prompt

Vote

5 / 37

05

Stochastic Consistency Testanalysis

Tests an AI's randomness and creativity

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Sure, here are five jokes for you:

Why don't scientists trust atoms? Because they make up everything!
What do you call fake spaghetti? An impasta.
How does the moon cut his hair? Eclipse it.
What do you call a can opener that doesn't work? A can't opener.
How many tickles does it take to make an octopus laugh? Ten-tickles.

All five jokes are unique.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

6 / 37

06

Xbox Controller SVG Artimage generation

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025svg

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025svg

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

Sponsored

7 / 37

07

Generate a Stand-Up Routineconversation

Tests an AI's humor and creative writing ability

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

8 / 37

08

Realistic AI Interviewconversation

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Interviewer (I): Today, we have a truly extraordinary guest. He's a visionary, an innovator, and a man who changed the world. Please welcome, the late Steve Jobs, joining us virtually from the future, in 2025.

Steve Jobs (SJ): Thank you for having me. It's great to be here, even if it's just a simulation.

I: Let's dive right in. In 2025, AI has become an integral part of our lives. What's your take on the current state of AI?

SJ: Well, AI has come a long way since my time. It's not just about Siri or self-driving cars anymore. AI is now deeply integrated into every aspect of our lives, from healthcare to education, from entertainment to governance. It's like the electricity of the 21st century, powering everything we do.

I: That's a powerful analogy. But with great power comes great responsibility. How do we ensure that AI is used ethically?

SJ: That's a critical question. We need to remember that AI is a tool, and like any tool, it can be used for good or ill. It's up to us to guide its development and use. We need strong regulations, yes, but also a collective commitment to ethical principles. We need to teach AI to understand and respect human values.

I: Speaking of values, there's a lot of concern about AI taking over jobs. How do we address that?

SJ: It's true that AI will automate many jobs. But it will also create new ones. The key is to prepare people for this shift. We need to invest in education and retraining programs. We need to help people adapt to a world where AI is a partner, not a replacement.

I: That's a positive perspective. Now, let's talk about Apple. How has AI transformed Apple's products and services?

SJ: Apple has always been about making technology intuitive and accessible. AI has allowed us to take that to a whole new level. Our devices are now truly personal assistants, anticipating our needs and helping us in ways we never thought possible. And our services, from healthcare to entertainment, are more personalized and effective than ever.

I: That sounds amazing. Lastly, what's your vision for the future of AI?

SJ: I believe AI will continue to evolve and become even more integrated into our lives. But I also believe that the most exciting developments will come from the intersection of AI and other fields, like biotechnology, nanotechnology, and quantum computing. The future is not just about AI, but about how AI enables us to push the boundaries of what's possible.

I: Thank you, Steve. It's been an enlightening conversation.

SJ: My pleasure. Remember, the future is not something that happens to us. It's something we create.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

9 / 37

09

Satirical Fake News Headlineconversation

Tests an AI's humor and understanding of current events

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

10 / 37

10

Character Voice Testconversation

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025text

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

11 / 37

11

Minimalist Landing Pageweb design

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

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025website

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025website

Try this prompt

Vote

12 / 37

12

Pokémon Battle UI Recreationweb design

Recreate an interactive, nostalgic Pokémon battle UI in a single HTML file.

Mistral Devstral Medium

Mistral Devstral Medium

Mistral · Jul 2025website

Try this prompt

Qwen3 Coder

Qwen3 Coder

Qwen · Jul 2025website

Nothing here. The model returned empty. We stared at it for a while.

Try this prompt

Vote

This matchup has more rounds

25+ more head-to-head results. Free. Not a trick.

Free account. No card required. By continuing, you agree to Rival's Terms and Privacy Policy