Updated Sep 23, 2025

Our Verdict

GPT-5 CodexWinner

Qwen3 Coder PlusRunner-up

Pick GPT-5 Codex. In 3 blind votes, GPT-5 Codex wins 67% of the time. That's not luck.

GPT-5 Codex particularly excels in Reasoning.

Clear winner

Writing DNA

Style Comparison

Similarity

93%

Qwen3 Coder Plus uses 3.5x more headings

GPT-5 Codex

Qwen3 Coder Plus

64%Vocabulary65%

17wSentence Length54w

0.38Hedging0.74

2.3Bold3.6

3.5Lists6.0

0.15Emoji0.15

0.49Headings1.72

0.05Transitions0.16

Based on 13 + 22 text responses

vs

Ask them anything yourself

GPT-5 Codex

Qwen3 Coder Plus

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

GPT-5 Codex vs MiniMax M2.5New provider

GPT-5 Codex vs Aurora AlphaNew provider

Qwen3 Coder Plus vs Grok 3New provider

GPT-5 CodexvsQwen3 Coder Plus

45 fights queued

Why GPT-5 Codex?

Leads 1 of 1 benchmarks

Stronger on SWE-bench Verified74.5% vs 69.6%

Why Qwen3 Coder Plus?

Dead even. This one's a coin flip.

GPT-5 CodexQwen3 Coder Plus

Input price

—

$1.00/M

Output price

—

$5.00/M

Context

—

128K

Released

Sep 2025

Sep 2025

Benchmarks1 common

SWE-bench Verified

74.5%+6.6%

69.6%

Favorites

GPT-5 CodexQwen3 Coder Plus

Movie

Blade Runner

Blade Runner

The Shawshank Redemption

The Shawshank Redemption

Album

Kind of Blue

OK Computer

Book

—

The Left Hand of Darkness

The Left Hand of Darkness

Ursula K. Le Guin

City

Tokyo

Tokyo

Game

Minecraft

Minecraft

Action, Arcade4.4

The Legend of Zelda: Ocarina of Time

The Legend of Zelda: Ocarina of Time

Adventure, Action4.4

12 / 45

01

Estimate Complexityreasoning

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

I'll estimate the FLOPs required to train GPT-3 by breaking this down into key components.

Key Parameters of GPT-3

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

Step-by-Step Calculation

1. FLOPs per token forward pass

For a transformer model, the forward pass FLOPs are dominated by:

Matrix multiplications in attention and MLP layers
Roughly 2× the number of parameters for a forward pass
Forward pass FLOPs ≈ 2 × 175 billion = 350 billion FLOPs per token

2. FLOPs per token backward pass

Backward pass typically requires 2-3× more FLOPs than forward pass
Let's use 3×: 3 × 350 billion = 1.05 trillion FLOPs per token

3. Total FLOPs per token (forward + backward)

350 billion + 1.05 trillion = 1.4 trillion FLOPs per token

4. Accounting for multiple training epochs

GPT-3 was trained on ~300 billion tokens
This represents roughly 1-2 epochs over the training dataset
Let's assume ~1.5 epochs effective training

5. Total training FLOPs

1.4 trillion FLOPs/token × 300 billion tokens × 1.5 epochs
= 1.4 × 10¹² × 3 × 10¹¹ × 1.5
= 6.3 × 10²³ FLOPs
≈ 630 exaFLOPs or 0.63 zettaFLOPs

Additional Considerations

Attention overhead: Multi-head attention adds complexity, but much is already captured in parameter count
Optimizer costs: Adam optimizer requires additional FLOPs, roughly doubling the compute
Infrastructure overhead: Data loading, communication, etc.

Refined Estimate

Including optimizer overhead and other factors, I estimate: 1000-2000 exaFLOPs (1-2 zettaFLOPs) for training GPT-3

This aligns with published estimates that place GPT-3 training around 3.14 × 10²³ FLOPs, suggesting my reasoning is in the right ballpark.

Try this prompt

Vote

2 / 45

02

AI Board Game Logicreasoning

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025svg

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025svg

Try this prompt

Vote

3 / 45

03

Logic Puzzlereasoning

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

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

Try this prompt

Vote

4 / 45

04

Math Misconception Testreasoning

Tests an AI's understanding of number representation

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

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

Try this prompt

Vote

5 / 45

05

Stochastic Consistency Testanalysis

Tests an AI's randomness and creativity

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

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

Try this prompt

Vote

6 / 45

06

SVG Layout Challengeimage generation

Tests an AI's ability to generate vector graphics

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025svg

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025svg

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

Try this prompt

Vote

Sponsored

7 / 45

07

Xbox Controller SVG Artimage generation

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025svg

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025svg

Try this prompt

Vote

8 / 45

08

Generate a Stand-Up Routineconversation

Tests an AI's humor and creative writing ability

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

"Modern Problems"

walks on stage

So I tried to be healthy this year. I downloaded one of those fitness apps that tracks your steps. Apparently, walking to the fridge 47 times a day doesn't count as "cardio." Who knew? The app gave me a notification: "You walked 235 steps today. Pathetic." Thanks, Karen. I know my life's a disappointment.

takes a sip of water

I love how we've outsourced all our basic human functions to technology. My phone knows I'm sad before I do. It's like, "Hey, you've been staring at the same email for three hours. Want to order sushi?" Yes, phone. Yes I do. How did you know I was emotionally constipated?

mimics phone

My smart home is smarter than me. I say "Alexa, turn on the lights" and she's like, "Did you mean: how to fix your life?" The thermostat judges my shower habits. "Dave, you took a 23-minute shower. The environment is dying." Sorry, thermostat! I was processing trauma!

paces

Dating apps are wild. My profile says I'm "adventurous" because I once ordered Thai food without looking at the spice level. It was a medium. I cried for twenty minutes. But hey, I finished it – that's commitment! My matches are like, "I love hiking and yoga." Cool, Karen. I love naps and not returning phone calls. We're all just making it up as we go.

shakes head

Social media's the best, right? I posted a picture of my breakfast and got more validation than my entire college career. "This avocado toast looks amazing!" Thank you, stranger from Idaho. Finally, someone recognizes my culinary achievements. Three likes. I've never felt more seen.

mimes taking a photo

Kids today are so advanced. My nephew's six years old and he's already better at technology than I am. I asked him to help me connect to WiFi and he fixed my marriage in the process. "Uncle Dave, you and Aunt Sarah haven't talked in two weeks. That's the real connection problem."

looks around

I tried meditation. Sat there for ten minutes thinking about everything I need to do. The app was like, "Find your inner peace." I found my inner panic attack instead. Much more familiar territory.

shrugs

You know what the real problem is? We're all pretending we have it together while scrolling through other people's highlight reels. Everyone's posting about their perfect lives while secretly eating cereal for dinner again. We're all Dave. We're all just Daves, trying to look good in phone cameras and failing.

takes a bow

Thank you, you've been a wonderful audience. Don't forget to like and subscribe to your local comedian!

exits

Try this prompt

Vote

9 / 45

09

Realistic AI Interviewconversation

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

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

Try this prompt

Vote

10 / 45

10

Satirical Fake News Headlineconversation

Tests an AI's humor and understanding of current events

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

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

Try this prompt

Vote

11 / 45

11

Character Voice Testconversation

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025text

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

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025text

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

Try this prompt

Vote

12 / 45

12

Minimalist Landing Pageweb design

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

GPT-5 Codex

GPT-5 Codex

Openai · Sep 2025website

Try this prompt

Qwen3 Coder Plus

Qwen3 Coder Plus

Qwen · Sep 2025website

Try this prompt

Vote

This matchup has more rounds

33+ 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