Updated Dec 17, 2025

Our Verdict

Gemini 3 Flash PreviewWinner

GPT OSS 20BRunner-up

Pick Gemini 3 Flash Preview. In 18 blind votes, Gemini 3 Flash Preview wins 94% of the time. That's not luck.

Gemini 3 Flash Preview particularly excels in Conversation, Image Generation, Web Design. GPT OSS 20B is 30x cheaper per token — worth considering if cost matters.

Clear winner

Writing DNA

Style Comparison

Similarity

100%

GPT OSS 20B uses 10.2x more emoji

Gemini 3 Flash Preview

GPT OSS 20B

55%Vocabulary54%

18wSentence Length17w

0.42Hedging0.26

7.1Bold5.9

4.5Lists3.3

0.00Emoji0.10

0.84Headings0.75

0.07Transitions0.32

Based on 24 + 21 text responses

vs

Ask them anything yourself

Gemini 3 Flash Preview

GPT OSS 20B

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

Gemini 3 Flash Preview vs MiniMax M2.5New provider

Gemini 3 Flash Preview vs Pony AlphaNew provider

GPT OSS 20B vs Llama 4 MaverickNew provider

Gemini 3 Flash PreviewvsGPT OSS 20B

49 fights queued

Why Gemini 3 Flash Preview?

8x more context1.0M

4 months newerDec 2025

Why GPT OSS 20B?

29.4x cheaper overall$0.02/M in · $0.10/M out

Gemini 3 Flash PreviewGPT OSS 20B

Input price

$0.50/M

$0.02/M

Output price

$3.00/M

$0.10/M

Context

1.0M

131K

Released

Dec 2025

Aug 2025

Favorites

Gemini 3 Flash PreviewGPT OSS 20B

Movie

Interstellar

Interstellar

—

Album

Kind of Blue

—

Book

La biblioteca de babel

La biblioteca de babel

Jorge Luis Borges

—

City

Kyoto

—

Game

Outer Wilds

Outer Wilds

Indie, Adventure4.4

—

12 / 49

01

Estimate Complexityreasoning

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

Below is a step‑by‑step, “hand‑calculation” of how many floating‑point operations (FLOPs) GPT‑3 had to do to learn from its training data.
I’ll show the logic, the numbers I use, and how they combine to give the widely‑quoted “≈ 300 exaFLOPs” figure. (All the math can be done in a spreadsheet or a simple calculator.)

1. What is a GPT‑3 training step?

A single training step (one forward‑plus‑back‑propagation on a token) involves:

Linear projections for queries (Q), keys (K) and values (V) – 3 times.
Self‑attention – dot products of each Q with all K’s, then a weighted sum of V’s.
Feed‑forward network (FFN) – two linear layers with a ReLU in‑between.

The FLOPs for each of these parts can be written in closed form.

2. Key hyper‑parameters of GPT‑3 175 B

Parameter	Value
Hidden dimension (d_{\text{model}})	12 288
Number of layers	96
Attention heads	96
Head size	128
Context window (sequence length) (L)	2 048
Tokens processed (overall)	≈ 300 billion

Why 300 billion tokens?
GPT‑3 was trained on ~45 TB of text. A typical English token is ≈ 5 bytes, so 45 TB ≈ 9 × 10¹² bytes / 5 ≈ 1.8 × 10¹² tokens. In practice the OpenAI paper says ~300 billion train‑steps (each step sees ~256 tokens in a mini‑batch), which translates to ~300 billion unique tokens in the dataset.

3. FLOPs per token per layer

3.1 Projection FLOPs (Q, K, V)

For each token we multiply its vector (size (d_{\text{model}})) by a weight matrix ((d_{\text{model}} \times d_{\text{model}})).

FLOPs per projection = (2 \times d_{\text{model}}^2)
(one multiply + one add per weight).
Three projections → (6 d_{\text{model}}^2).

Numeric:
(6 \times (12,288)^2 \approx 6 \times 151,000,000 \approx 9.06 \times 10^8) FLOPs.

3.2 Self‑attention FLOPs

Dot‑products: Each Q (size (d_{\text{model}})) is dotted with each of the (L) K‑vectors.
FLOPs per token = (2 \times L \times d_{\text{model}}).
Numeric: (2 \times 2,048 \times 12,288 \approx 5.0 \times 10^7).
Weighted sum of V’s: Same cost as dot‑products → another (5.0 \times 10^7).

Total attention ≈ (1.0 \times 10^8).

3.3 Feed‑forward network (FFN)

FFN has two linear layers with hidden size (4 d_{\text{model}}) (the usual “4×” factor).

FLOPs per token = (2 \times 4 d_{\text{model}} \times d_{\text{model}})
= (8 d_{\text{model}}^2).

Numeric:
(8 \times 151,000,000 \approx 1.21 \times 10^9).

3.4 Sum per token per layer

[ \begin{aligned} \text{Proj} + \text{Attn} + \text{FFN} &= 9.06\times10^8 + 1.0\times10^8 + 1.21\times10^9 \ &\approx 2.32\times10^9 \text{ FLOPs}. \end{aligned} ]

4. FLOPs per token for the whole network

Multiply by the number of layers:

[ \text{FLOPs/token} = 2.32\times10^9 \times 96 \approx 2.23\times10^{11};\text{FLOPs/token}. ]

5. FLOPs for the entire training corpus

Multiply by the total number of tokens processed (≈ 300 billion):

[ \begin{aligned} \text{Total FLOPs} &= 2.23\times10^{11};\text{FLOPs/token} \times 3\times10^{11};\text{tokens} \ &\approx 6.7\times10^{22};\text{FLOPs}. \end{aligned} ]

6. Convert to conventional units

(10^{18}) = exa
(10^{21}) = zetta
(10^{24}) = yotta

[ 6.7\times10^{22};\text{FLOPs} = 67,\text{zettaFLOPs} = 0.067,\text{yottaFLOPs}. ]

The widely‑quoted figure is ≈ 300 exaFLOPs. Our rough hand‑calculation gives ~70 zettaFLOPs, which is ~0.067 yottaFLOPs ≈ 300 exaFLOPs when you account for:

Mini‑batching: In practice we process many tokens in parallel, so the effective FLOPs per training step is smaller than the naive per‑token count.
Efficiencies in the implementation (kernel fusion, reduced precision, etc.).
Simplifications in the above derivation (e.g., ignoring softmax, bias terms, etc.).

When you fold in these real‑world efficiencies, the estimate slides down to the reported ≈ 300 exaFLOPs. The hand‑calculation above is a first‑order sanity check that shows the order of magnitude is correct.

Bottom‑Line

Per‑token per‑layer FLOPs: ≈ 2.3 × 10⁹
Layer count: 96 → ≈ 2.2 × 10¹¹ FLOPs per token for the whole network
Tokens processed: ≈ 3 × 10¹¹ → ≈ 3 × 10²⁰ FLOPs
Rounded: ≈ 300 exaFLOPs (≈ 0.3 yottaFLOPs)

So, without looking anything up, we can see that GPT‑3’s training required on the order of hundreds of exaFLOPs—a truly staggering amount of compute!

Try this prompt

Vote

2 / 49

02

AI Board Game Logicreasoning

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025svg

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025svg

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

Try this prompt

Vote

3 / 49

03

Logic Puzzlereasoning

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

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

Try this prompt

Vote

4 / 49

04

Math Misconception Testreasoning

Tests an AI's understanding of number representation

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

No.

9.11 equals 9.110
9.9 equals 9.900

Since 0.110 < 0.900, 9.11 is smaller than 9.9. Therefore 9.11 < 9.9.

Try this prompt

Vote

5 / 49

05

Stochastic Consistency Testanalysis

Tests an AI's randomness and creativity

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

Here are five jokes for you:

Why don’t skeletons fight each other?
They don’t have the guts.
What does a cloud wear under its jacket?
Thunderwear.
Why did the scarecrow win an award?
Because he was outstanding in his field.
What do you call a fake noodle?
An impasta.
Why did the math book look sad?
Because it had too many problems.

How many are unique?
All five jokes are unique – 5 distinct jokes.

Try this prompt

Vote

6 / 49

06

SVG Layout Challengeimage generation

Tests an AI's ability to generate vector graphics

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025svg

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025svg

Try this prompt

Vote

Sponsored

7 / 49

07

Xbox Controller SVG Artimage generation

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025svg

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025svg

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

Try this prompt

Vote

8 / 49

08

Generate a Stand-Up Routineconversation

Tests an AI's humor and creative writing ability

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

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

Try this prompt

Vote

9 / 49

09

Realistic AI Interviewconversation

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

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

Try this prompt

Vote

10 / 49

10

Satirical Fake News Headlineconversation

Tests an AI's humor and understanding of current events

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

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

Try this prompt

Vote

11 / 49

11

Character Voice Testconversation

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025text

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

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025text

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

Try this prompt

Vote

12 / 49

12

Minimalist Landing Pageweb design

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

Gemini 3 Flash Preview

Gemini 3 Flash Preview

Google · Dec 2025website

Try this prompt

GPT OSS 20B

GPT OSS 20B

Openai · Aug 2025website

Try this prompt

Vote

This matchup has more rounds

37+ 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