ElevenLabs Turbo v2.5

Overview

This model converts text input into audio output across 32 languages. It operates as a fast, cheap-tier model with strong audio quality, delivering quicker generation than the Multilingual variant while maintaining higher fidelity than the Flash variant. On Kyma, the model runs through an OpenAI-compatible endpoint using a single API key. It supports prompt caching, which bills repeated prompt prefixes at this model’s cached input rate. New accounts include a $0.50 free credit with no card required. Every request receives automatic failover, and responses return exact billing data in usage.cost alongside an X-Kyma-Model header to confirm routing. The model accepts a 5000-token context window and is strictly designed for speech generation. It does not support reasoning, vision, structured outputs, or multimodal inputs, and it will not handle complex logical tasks or data extraction.

Specs

Field	Value
Model ID	`eleven-turbo-v2-5`
Best for	Default balanced voice for medium-latency narration, podcasts
Context window	5,000
Input modalities	Text
Output modalities	Audio
Tool calling	Yes
Structured outputs	Yes
Prompt caching	Yes
Speed	Fast
Cost band	Cheap
Release stage	Stable

Pricing

	Per 1M tokens
Input	$0.20
Output	$0.60

Use this when

Podcast narration generation — Produces consistent voiceovers for long-form audio content with balanced latency.
Multilingual audiobook creation — Converts text to speech across 32 supported languages without sacrificing audio clarity.
Interactive voice applications — Delivers fast audio responses for conversational agents where moderate latency is acceptable.
E-learning module voiceovers — Generates clear instructional audio at a lower cost than premium TTS alternatives.

Not ideal for

Do not use this model for real-time conversational voice agents requiring sub-100ms latency or for tasks requiring structured data extraction and logical reasoning.

Pick something else when

You need maximum audio fidelity regardless of cost or speed: use eleven-v3.
You require the fastest possible generation for high-volume text: use eleven-flash-v2-5.
You need broader language coverage beyond the supported 32: use eleven-multilingual-v2.

Example

from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="ky-...")

response = client.chat.completions.create(
    model="eleven-turbo-v2-5",
    messages=[{"role": "user", "content": "..."}],
)

Agent query example

Ask the API which models fit, instead of hardcoding an id:

curl "https://kymaapi.com/v1/models?recommended_for=generate-speech&tools=true&quality_tier=strong"

FAQ

Does this model support prompt caching on Kyma? Yes. Repeated prompt prefixes are billed at this model’s cached input rate, reducing costs for recurring text segments. Can I use this model for structured outputs or reasoning tasks? No. It is strictly a text-to-speech model and does not support reasoning, vision, or structured JSON outputs. How do I verify which model actually processed my request? Check the X-Kyma-Model response header, which reports the exact model ID that ran, and review the usage.cost field for precise billing details.

​Overview

​Specs

​Pricing

​Use this when

​Not ideal for

​Pick something else when

​Example

​Agent query example

​FAQ