MiniMax Speech 2.8 Turbo
Save 20%Officialminimax/speech-2.8-turbo
The low-latency tier of MiniMax text-to-speech, built for voice agents and real-time narration where time-to-first-audio matters.
- Input modalities
- Text
- Output modalities
- Audio
- Upstream release
- 2026-01-23
- RouteMux name
- minimax/speech-2.8-turbo
What is MiniMax Speech 2.8 Turbo
Speech 2.8 Turbo (model ID speech-2.8-turbo) is the speed-first tier of MiniMax's Speech 2.8 family, released together with Speech 2.8 HD on January 23, 2026. MiniMax pitches it on generation speed with natural-sounding delivery, and its pricing page is more specific: the same capabilities as HD, optimized for speed and cost.
In practice that means everything new in Speech 2.8 is here too. Write sound tags such as (laughs), (breath) or (sighs) into the text and the model voices them; there are 19, and MiniMax's docs limit them to the two Speech 2.8 tiers. The 40 languages MiniMax lists are the same as well. In MiniMax's framing the tiers differ only in what they optimize: Turbo for speed and cost, HD for audio quality and emotional range. If the finer detail matters more, moving to HD is a change to the model field. MiniMax does not publish latency figures for Turbo.
On RouteMux it is the OpenAI-compatible /v1/audio/speech endpoint, returning MP3 bytes synchronously, so read the response as binary. voice is required; a request without it gets a 400 before synthesis and is not billed. speed ranges from 0.5 to 2. response_format accepts mp3 or nothing: wav, opus or flac return a 400 instead of quietly coming back as MP3. Each request's input must be under 10,000 characters.
Billing counts billable characters in your input: a Chinese character counts as 2, and letters, punctuation and spaces count as 1 each.
MiniMax Speech 2.8 Turbo API pricing
Key capabilities
- Tuned for speed and cost
- MiniMax positions it as the same capabilities as HD, generated faster and for less.
- The same 19 sound tags
- (laughs), (breath), (sighs), (gasps) and the rest work in the text as they do on HD.
- Synchronous MP3
- One request returns the finished audio bytes — no job to create, nothing to poll.
- 20% below official pricing
- Billed per input character, with Chinese characters counting as 2. The pricing section above has the rate.
Limited time10% extra Credits on your first top-up — up to 200 bonus Credits
Applies to your first top-up within 72 hours of signing up. Minimum $20, once per account.
Interface and protocols
Use any of these IDs to call this model via the API.
minimax/speech-2.8-turboReplace the ROUTEMUX_KEY placeholder with your API key. Create one →
from openai import OpenAI
client = OpenAI(
base_url="https://api.routemux.com/v1",
api_key="<ROUTEMUX_KEY>",
)
completion = client.chat.completions.create(
model="minimax/speech-2.8-turbo",
messages=[{"role": "user", "content": "What is the meaning of life?"}],
)
print(completion.choices[0].message.content)Performance (last 30 days)
MiniMax Speech 2.8 Turbo FAQ
›What is MiniMax Speech 2.8 Turbo?
The low-latency tier of MiniMax text-to-speech, built for voice agents and real-time narration where time-to-first-audio matters.
›How much does the MiniMax Speech 2.8 Turbo API cost?
Through RouteMux, MiniMax Speech 2.8 Turbo is priced at $48, below the official list price.
›How do I call MiniMax Speech 2.8 Turbo via API?
MiniMax Speech 2.8 Turbo is OpenAI-compatible: point your base URL at RouteMux and set the model field to minimax/speech-2.8-turbo — no code changes needed.
›Which input and output modalities does MiniMax Speech 2.8 Turbo support?
MiniMax Speech 2.8 Turbo accepts text as input and produces audio as output.
›When was MiniMax Speech 2.8 Turbo released?
MiniMax Speech 2.8 Turbo was released on January 23, 2026.
Sources
Reviewed by the RouteMux team on 2026-10-24