Side by side

See the difference.

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

Scroll to compare models →

Comparison of Google: Gemini 3.8 Flash Lite TTS, abr-ai/niagara-84m-batch.en
2 modelsgoogleGoogle: Gemini 3.8 Flash Lite TTSRemove ×abr-aiabr-ai/niagara-84m-batch.enRemove ×
What’s differentSeparate spoken text from delivery instructions, or script a two-voice dialogue with per-turn style overrides.0.084B parameters · 1 language · Encoder: State-space model · Decoder: CTC
DateSep 23, 2026ReleasedOct 17, 2026Discovered
Listed rate$0.03/ 1K charactersfal · all routes ↗See provider quotes
Provider endpoints1 endpoint →0 endpoints →
AA Native Voice ↗Unversioned · provider native voices1,251.27 ↗Provider native voices · all categories · US + UK accents · 8 listed voicesIndependent result95% CI 1,238.27–1,264.27 · n=3,025Not reported
AMI-Cleaned WER ↗English short-form; normalization unversionedNot reported12.01% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
Earnings22-Cleaned-AA-chunked WER ↗English short-form; normalization unversionedNot reported10.83% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
Gigaspeech-Cleaned WER ↗English short-form; normalization unversionedNot reported10.41% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
LS Clean WER ↗English short-form; normalization unversionedNot reported3.09% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
LS Other WER ↗English short-form; normalization unversionedNot reported7.6% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
SPGISpeech WER ↗English short-form; normalization unversionedNot reported2.61% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
URGENT2024 WER ↗English short-form; normalization unversionedNot reported20.16% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
Voice Arena Monsoon WER ↗English short-form; normalization unversionedNot reported8.82% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
Voxpopuli-AA-Cleaned WER ↗English short-form; normalization unversionedNot reported4.33% WER ↗abr-ai/niagara-84m-batch.enLeaderboard entry
Context window32.8K—
Max output16.4K—
Parameters—0.084B
Audio inputNot reportedSupported
TranscriptionNot reportedSupported

Add another model

Search all models →