Exa API (formerly Metaphor) vs Perplexity API (Sonar models): what AI actually recommends
The same buying questions, run after run, to ChatGPT, Claude, Gemini, and Perplexity: the record of which of these two they name first. No opinions, no scores we invented: what the models said.
AI currently favors Perplexity API (Sonar models). Ahead in 1 of 1 shared category on consensus rank.
That is the latest close, not a verdict for all time. The model-by-model board below shows where they agree, where they split, and how both brands moved.
At the latest close, Exa API (formerly Metaphor) leads 0 and Perplexity API (Sonar models) leads 1 across the 1 category they both appear in.
Beyond this pairing, Exa API (formerly Metaphor) is ranked in 1 category; Perplexity API (Sonar models) is ranked in 1 category. The board below is the shared slice, where the four models place these two side by side.
The record, category by category.
Consensus rank · latest close- #5Perplexity API (Sonar models)New entry at the last closespread 10Gemini#2
- #7Exa API (formerly Metaphor)New entry at the last closespread 10Gemini#3
On consensus rank, Perplexity API (Sonar models) leads here, #5 to Exa API (formerly Metaphor)'s #7. 21 to 18 on the 0-100 score. Every model that separates the two ranks Perplexity API (Sonar models) ahead. Perplexity API (Sonar models) is new to the board at this close.
What the models actually said.
The ranking is the answer; these are the words behind it. Every quote is from a recorded model answer — follow the category to read it whole.
Exa API (formerly Metaphor)
Exa API (formerly Metaphor): Uses neural search rather than keyword search to understand the semantic meaning of queries, making it exceptionally powerful for LLMs needing deep, relevant links.
Gemini
Perplexity API (Sonar models)
Perplexity API (Sonar models): Provides state-of-the-art conversational search capabilities natively grounded in real-time web data with built-in citation support.
Gemini
Where the models split.
Across the 1 category Exa API (formerly Metaphor) and Perplexity API (Sonar models) both appear in, the models never break ranks: in each one, the same brand leads on ChatGPT, Claude, Gemini, and Perplexity alike.
No model runs against the grain here: each of the four ranks Perplexity API (Sonar models) ahead at least as often as not, with Gemini the most lopsided.
- ChatGPTsplit 0-0
- Claudesplit 0-0
- Geminifavors Perplexity API (Sonar models) 1-0
- Perplexitysplit 0-0