MiniMax Speech
MiniMax · China · Speech & audio
China Closed sourceIndependent test
Showing version: Speech 2.8（HD / Turbo） · 2026-01-23 · Version lineage ↓
01
Overview
MiniMax's text-to-speech model line focused on multilingual synthesis and high-fidelity voice cloning: starting with Speech-02, expanding to 40+ languages with 2.5, and currently led by Speech 2.8, served via the MiniMax open platform and Hailuo AI.
Release date2026-01-23
02 / Provenance
Field checks and source notes
When sources disagree on a field, every reading is listed here with its own citation. We do not take a midpoint, invent a missing figure, or decide which side is right — follow the citation and judge for your case.
These field-level notes are recorded verbatim in Chinese in the source data. Open the Chinese version
03 / Plain layer
Plain-language benchmarks
Naturalness 93 · Human preference
Accuracy (WER) Not disclosed
Speed / latency Not disclosed
Languages supported Not disclosed
Price / free Not disclosed
04
Practical facts
| Context length | Not disclosed |
|---|---|
| Max output | Not disclosed |
| Parameters | Not disclosed |
| Input | Not disclosed |
| Output price | Not disclosed |
| CNY | Not disclosed |
| Free tier | Not disclosed |
| Output speed | Not disclosed |
| Release date | |
| License | Not disclosed |
Show technical data
Technical benchmark table
| Benchmark | Score | Max / unit | Source type | Source | Test date | Conditions |
|---|---|---|---|---|---|---|
| TTS preference tts-preference | 93 | elo_like | Independent test | Humanness Index 综合分（0-100 量纲，非 Arena Elo）；该页同时记录 Speech 2.8 中位 TTFB 约 325ms |
| Version | Release date | Status | Openness | Parameters | What changed | Detail | |
|---|---|---|---|---|---|---|---|
| Speech 2.8（HD / Turbo）Showing version | Active | Closed source | Not disclosed | Current flagship generation released January 2026 (third-party model trackers record Jan 23; official docs list Speech-2.8), supporting 40+ languages and natural sound tags. | Detail | ||
| Speech 2.6（HD / Turbo） | Superseded | Closed source | Not disclosed | Released late October 2025 for low-latency voice-agent workflows. | Detail | ||
| Speech 2.5（HD / Turbo） | Superseded | Closed source | Not disclosed | Official launch: improved multilingual expressiveness and voice-cloning fidelity, 40+ languages. | Detail | ||
| Speech-02（HD / Turbo） | Superseded | Closed source | Not disclosed | Start of the Speech line (official release-notes index records 2025-04-02), in HD and Turbo tiers for multilingual TTS and cloning. | Detail |
Last updated: · Sources on detail page
06 / Compare