Perplexity vs Gemini

As of 2026-07-18, Perplexity (Perplexity: Sonar Pro, 7.7/10) leads Gemini (Google: Gemini 3.5 Flash, 7.2/10) in our daily benchmark, where over 350 AI models are evaluated.

Both can search the web while answering, which is exactly why they are worth comparing: for research, a model with sources is worth more than a model with a good memory.

Perplexity

Perplexity: Sonar Pro

Score: 7.7/10

Language: 9.0/10

Speed: 68 t/s

$3.00

Google

Gemini

Google: Gemini 3.5 Flash

Score: 7.2/10

Language: 8.0/10

Speed: 419 t/s

$1.50

Side by side

	Perplexity	Gemini
In today’s benchmark	Perplexity: Sonar Pro	Google: Gemini 3.5 Flash
Score	7.7/10	7.2/10
Language	9.0/10	8.0/10
Speed	68 t/s	419 t/s
Input price/1M	$3.00	$1.50

What separates them?

Perplexity is built around search from the ground up: the answer arrives with sources you can click through to. Gemini is a general model with search as one capability among many, but with longer context and broader use. For pure research Perplexity is often the more practical tool; for everything else Gemini is more versatile.

The sources are the point

Whichever you choose: open the sources. Both can misrepresent a source, and both can link to something that does not say what they claim. The value is in having the sources served to you — not in being spared from reading them.

hvilkenAI is independent. We have no affiliate agreements, sponsored placements or commercial ties to the AI providers we test. The numbers above come from today’s benchmark.

Frequently asked questions

Which is best for research?

See the ranking above, updated daily. Note that the language score measures language quality, while the value in research also lies in how sources are handled.

Is Perplexity free?

Perplexity has a free tier with limits and a paid tier with access to stronger models.

Can they invent sources?

Models with real search do so far less often than models without. But always check that the source actually says what is claimed.