Comparing AI models in n8n in Spreadsheet

Comparing AI Models: Chat GPT, Haiku, Sonnet, Free V3, and Gemini Flash

Today we're going to compare several AI models to see how they stack up against each other in quality, speed, and cost-effectiveness.

The Contenders

I'll be testing five different models:

Deep Seek V3

Sonet

4o Mini

Gemini Flash

Haiku

Testing Methodology

For each model, I'll measure:

Time to complete

Tokens in/out

Quality rating (my subjective assessment)

Cost

Deep Seek V3 Results

I've already tested Deep Seek V3, and the quality was pretty decent. The formatting was actually correct, which is a plus point.

Quality rating: 4/5 Note: It didn't provide a hook that was requested in the system prompt.

4o Mini Results

Time: ~4 seconds (incredibly fast!) Cost: 0.006 Tokens: [Not specified in transcript]

The formatting with 4o Mini was really bad - everything was crumbled together. I had to ask it to fix the formatting.

Quality rating: 3/5

Gemini Flash Results

Time: ~3 seconds Cost: [Not specified in transcript] Tokens: [Not specified in transcript]

Gemini Flash provided a clear step-by-step guide with proper keywords like "connecting to DNS" and "adding your droplet to domain." The time codes looked correct, and it included a hook.

Quality rating: 4/5

Haiku Results

Time: ~9 seconds Cost: Higher than others

Haiku surprisingly underperformed. It included "Pro tips" despite my system prompt specifically requesting not to include them. The structure wasn't as clear as some of the other models.

Quality rating: 3/5

Sonet Results

Time: ~10 seconds Cost: Expensive compared to the others

The formatting was off, which was disappointing. Based on my previous experience, Sonet usually outperforms other models, but not in this test.

Quality rating: 3/5

The Winner: Gemini Flash

After comparing all five models, Gemini Flash comes out on top with:

Fast response time (~3 seconds)

Free to use

Best quality output

Proper formatting and structure

Deep Seek V3 also performed well with correct formatting.

FAQ

Q: Are there rate limits on Gemini Flash FREE tier?

A: Yes, 15 requests per minute (RPM), paid version 2000 RPM https://ai.google.dev/gemini-api/docs/rate-limits

Q: Which model had the best formatting?

A: Deep Seek V3 had the most correct formatting based on the system prompt.

Q: Which model was the most cost-effective?

A: Gemini Flash offers the best balance of quality and cost (free).

Q: Which model was the fastest?

A: Gemini Flash was the fastest at approximately 3 seconds, with 4o Mini close behind at 4 seconds.