← All models

ElevenLabs Flash v2.5

Audio Available Full comparison ↗

elevenlabs/flash-v2-5 · by ElevenLabs · neural-tts

Pricing — 1 offering(s)

ElevenLabs

About this offering: ElevenLabs' low-latency text-to-speech model for real-time and conversational use: about 3x lower latency than Multilingual v2, at some cost in voice quality, across 32 languages. The pay-as-you-go API rate is $0.04 per 1K characters ($40 per 1M), half Multilingual v2's $80 (ElevenLabs API pricing, checked 2026-10-11).

Text characters

  • $60.00 / 1M characters Historical 2024-06-01 → 2026-07-29
  • $50.00 / 1M characters Historical 2026-07-30 → 2026-09-29
  • $40.00 / 1M characters Current 2026-09-30 → present
    ElevenLabs' pay-as-you-go API rate for its Flash / Turbo models: $0.04 per 1K characters ($40 per 1M), down…

    ElevenLabs' pay-as-you-go API rate for its Flash / Turbo models: $0.04 per 1K characters ($40 per 1M), down from $0.05 per 1K when checked on 2026-07-30. ElevenLabs doesn't publish the date the rate changed.

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

See how ElevenLabs Flash v2.5 fits into a cost-aware routing setup

See how →

Capability profile

How ElevenLabs Flash v2.5 rates across core capability dimensions, with the task-level evidence behind each rating.

voice naturalness Moderate

Strong naturalness for a latency-optimised model; noticeable quality gap vs Multilingual v2 on close listening.

language support Strong

32 languages; slightly broader language coverage than Multilingual v2.

voice variety Strong

Access to the full ElevenLabs voice library and voice cloning.

streaming latency Strong

~75ms first-byte latency; designed for real-time conversational AI.

cloning support Strong

Full voice cloning, speech-to-speech, emotion control via API.

Benchmarks

Benchmark Score Config Source
First-byte latency (vendor-reported) 75 ms Streaming endpoint, ElevenLabs infrastructure. Vendor-reported, not independently verified. source ↗
MOS (naturalness) — No formally published MOS score. Community comparisons rate it below Multilingual v2 but above commodity TTS APIs. —

Operator guidance

The default ElevenLabs choice for interactive and real-time applications. At ~$50/1M chars (half the cost of Multilingual v2), it covers most use cases. Upgrade to Multilingual v2 only when audio quality is the primary differentiator (e.g. final production narration).

Use cases

Limitations

Citations