Back to models
Perplexity logo

Perplexity

Sonar Deep Research

Perplexity’s deepest research tier with citation and search-query pricing components.

Overall score
86
deep-researchcitationssearch
Context window
Search-context dependent
Speed
Deliberate
Input pricing
$2 / 1M input tokens
Output pricing
$8 / 1M output tokens

Score breakdown

Capability89
Use-case fit94
Cost efficiency73
Speed70
Reliability87
Agent readiness81
Ecosystem84

Scores combine benchmark signals, product experience, and editorial weighting. Use them as a practical guide, not an absolute truth claim.

Best for

Research

Works with

deep research workflowscitation-heavy reportsanalyst assistants

Modalities

textsearch

Strengths

  • Built specifically for deeper research and citation-heavy output.
  • Makes source-backed reporting more explicit than generic chat models.

Things to watch

  • The pricing model has multiple moving parts and needs careful review.
  • Latency and cost profile are less suitable for lightweight interactive use.

Best for