Aller au contenu

MiniMax Speech 2.8 HD MiniMax · text-to-speech (TTS) API

1,350 ms time to first audio on Speko's board (read October 22, 2026); not measured by Coval, so no speed tier.

Benchmark summary: No speed tier: Coval, the source of the tiers, does not measure this model.

MiniMax Speech 2.8 HD is a proprietary text-to-speech (TTS) API from MiniMax, measured by two independent benchmarks: Artificial Analysis and Speko.

Speed tier

Not measured by Coval

No speed tier: Coval, the source of the tiers, does not measure this model.

Median time to first audioCoval, 30 days
—
P95 time to first audio (slow turns)Coval, 30 days
—
Naturalness EloSpeko
1431
Listener Elo (95% interval)Artificial Analysis
1,174 ±10
Numbers and dates (0 to 1)Speko
0.15
List price per 1M charactersall sources
$100
  • Customer service (Artificial Analysis): 30th, Elo 1170 ±24

Data read October 22, 2026. No combined score: methodology

Gradium TTS vs MiniMax Speech 2.8 HD Fastest TTS APIs, by speed tier

Speed

Speko's board plots it at 1,350 ms time to first audio from US East; its model page reports a 294 ms sweep.

Voice quality

Speko rates its naturalness at 1,431 Elo from blind A/B votes on phone-agent lines (33rd highest of the 43 models Speko rates). In the Artificial Analysis Speech Arena it holds 1,174 ±10 Elo (95% interval), 24th highest of the 98 models Artificial Analysis rates; scores whose intervals overlap are a tie. In its Customer Service category, Artificial Analysis places it 30th, at 1,170 Elo.

Numbers and dates

Speko's pronunciation test on numbers, dates and currency scores it 0.15 on a 0-to-1 scale (29th highest of the 30 models Speko scores). This is the accuracy measure used here; Coval's word error rate separates models too little to compare them.

Price

Artificial Analysis lists $100. Speko lists about $100. List prices, as each source prints them.

Languages tested

Tested in Spanish, German, French, Portuguese, Japanese, Chinese, Arabic, Hindi, Vietnamese, Filipino and Norwegian by Artificial Analysis or Speko. A language missing here was not tested, which does not mean it is unsupported.

Benchmark measurements

Speko

P50 TTFA, US East
294 ms
Naturalness (Elo)
1,431
Head-to-head win rate
36%
Pronunciation robustness (0 to 1)
0.15
Voice drift (lower is steadier)
12
Price per 1M characters
100 USD
Measured on
2026-07-03
Speko model
speech-2.8-hd
Speko model page
benchmarks.speko.ai
Time to first audio, board chart
1,350 ms

Artificial Analysis Speech Arena

Customer service rank
30
Customer service Elo
1,170
Customer service Elo 95% interval (±)
24
Speech Arena Elo (blind listener preference)
1,174
Elo 95% interval (±)
10
Arena rank
24
Arena rank range
21-25
Arena appearances
5,194
Voices tested
8
Price per 1M characters
100 USD
Released
Feb 2026
Artificial Analysis model
Speech 2.8 HD
Artificial Analysis leaderboard
artificialanalysis.ai

Price (all sources)

Lowest list price per 1M characters
100 USD
Highest list price per 1M characters
100 USD

Not collected

Similar TTS models