Side by side

See the difference.

Scores, capabilities and costs for the models you’re considering.

Scroll to compare models →

Comparison of Perplexity: Sonar Reasoning Pro, AssemblyAI: Universal-3.5 Pro
2 modelsperplexityPerplexity: Sonar Reasoning ProRemove ×assemblyaiAssemblyAI: Universal-3.5 ProRemove ×
What’s differentNote: Sonar Pro pricing includes Perplexity search pricing. See details here Sonar Reasoning Pro is a premier reasoning model powered by DeepSeek R1 with Chain of Thought (CoT). Designed for...Universal-3.5 Pro is AssemblyAI's speech-to-text model served through its Sync API, returning a complete transcript with word-level timestamps in a single synchronous response for audio clips up to 120…
DateMar 7, 2025ListedSep 22, 2026Listed
Listed rate$2.00 in / $8.00 out/ 1M tokensPerplexity · via OpenRouter · all routes ↗perplexity$83.3333 in / $0.00 out/ 1M tokensOpenRouter · all routes ↗
Provider endpoints2 endpoints →1 endpoint →
AMI-Cleaned WER ↗English short-form; normalization unversionedNot reported10.63% WER ↗assemblyai/universal-3-5-proLeaderboard entry
Earnings22-Cleaned-AA-chunked WER ↗English short-form; normalization unversionedNot reported6.05% WER ↗assemblyai/universal-3-5-proLeaderboard entry
Gigaspeech-Cleaned WER ↗English short-form; normalization unversionedNot reported7.62% WER ↗assemblyai/universal-3-5-proLeaderboard entry
LS Clean WER ↗English short-form; normalization unversionedNot reported1.1% WER ↗assemblyai/universal-3-5-proLeaderboard entry
LS Other WER ↗English short-form; normalization unversionedNot reported2.11% WER ↗assemblyai/universal-3-5-proLeaderboard entry
SPGISpeech WER ↗English short-form; normalization unversionedNot reported1.81% WER ↗assemblyai/universal-3-5-proLeaderboard entry
URGENT2024 WER ↗English short-form; normalization unversionedNot reported9.08% WER ↗assemblyai/universal-3-5-proLeaderboard entry
Voice Arena Monsoon WER ↗English short-form; normalization unversionedNot reported3.49% WER ↗assemblyai/universal-3-5-proLeaderboard entry
Voxpopuli-AA-Cleaned WER ↗English short-form; normalization unversionedNot reported1.92% WER ↗assemblyai/universal-3-5-proLeaderboard entry
Context window128K—
Max output115.2K—
Parameters——
Image inputSupportedNot reported
Audio inputNot reportedSupported
TranscriptionNot reportedSupported
ReasoningSupportedNot reported

Add another model

Search all models →