Aller au contenu

Gradium TTS Gradium · text-to-speech (TTS) API

Headroom

55 ms median and 101 ms P95 time to first audio on Coval (30 days to October 22, 2026): Headroom speed tier.

Benchmark summary: Headroom speed tier on Coval's P95 (headroom to spend elsewhere in the turn), a tie within 10 ms with Alibaba Qwen3 TTS Fast and Inworld Realtime TTS-2 Flash on slow turns. On the Speko speed and naturalness frontier and the Coval and Artificial Analysis frontier: no other model is both faster and rated higher.

Gradium TTS is a proprietary text-to-speech (TTS) API from Gradium, measured by three independent benchmarks: Artificial Analysis, Coval and Speko.

Speed tier

Headroom

P95 at or under 101 ms. Headroom to spend elsewhere in the turn.

Median time to first audioCoval, 30 days
55 ms
P95 time to first audio (slow turns)Coval, 30 days
101 ms
Naturalness EloSpeko
1592
Listener Elo (95% interval)Artificial Analysis
1,176 ±13
Numbers and dates (0 to 1)Speko
0.27
List price per 1M charactersall sources
$35.9 to $57.8

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

Fastest TTS APIs, by speed tier

Speed

On Coval's 30-day window ending October 22, 2026 (3,376 runs from US East), Gradium TTS starts speaking after 54.5 ms on a typical turn (median) and 100.7 ms on a slow turn (P95, the slowest one in twenty). That P95 is 4th lowest of the 32 models Coval measures, which puts it in the Headroom tier: headroom to spend elsewhere in the turn. On the median it is a tie (gap under 10 ms) with Alibaba Qwen3 TTS Fast and Inworld Realtime TTS-2 Flash. Speko's board plots it at 55 ms time to first audio from US East; its model page reports a 105 ms sweep.

Voice quality

Speko rates its naturalness at 1,592 Elo from blind A/B votes on phone-agent lines (6th highest of the 43 models Speko rates). In the Artificial Analysis Speech Arena it holds 1,176 ±13 Elo (95% interval), 23rd highest of the 98 models Artificial Analysis rates; scores whose intervals overlap are a tie. In its Customer Service category, Artificial Analysis places it 16th, at 1,242 Elo.

Numbers and dates

Speko's pronunciation test on numbers, dates and currency scores it 0.27 on a 0-to-1 scale (18th 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 $47.8. Speko lists about $57.8. Gradium's own plans work out to $57.8 (XS), $47.8 (S), $37.8 (M), $35.9 (L) per 1M characters (gradium.ai/pricing). List prices, as each source prints them: a range of $35.9 to $57.8.

Languages tested

Tested in Spanish, German, French and Portuguese by Artificial Analysis or Speko. A language missing here was not tested, which does not mean it is unsupported.

Benchmark measurements

Latency and accuracy — Coval (30-day window)

Mean TTFA
62.6 ms
P50 / median time-to-first-audio (TTFA)
54.5 ms
P90 TTFA
89.3 ms
P95 TTFA
100.7 ms
Latency consistency (standard deviation)
20.1 ms
Leading silence
3 ms
Time to first byte (network arrival)
59.6 ms
Accuracy: word error rate (WER)
2.2%
WER basis
pooled
Latency runs
3,376 runs
Accuracy runs
3,374 runs
Window
30d
Coval snapshot
2026-10-22T14:00:19.256741Z
Coval model ID
gradium-tts-beta-202609
Coval model page
benchmarks.coval.ai
Tagged multilingual
✓
Feature tags
multilingual, voice-cloning

Speko

P50 TTFA, US East
105 ms
Naturalness (Elo)
1,592
Pronunciation robustness (0 to 1)
0.27
Voice drift (lower is steadier)
14
Price per 1M characters
57.8 USD
Measured on
2026-10-01
Speko model
Gradium TTS
Speko model page
benchmarks.speko.ai
Time to first audio, board chart
55 ms

Artificial Analysis Speech Arena

Customer service rank
16
Customer service Elo
1,242
Customer service Elo 95% interval (±)
27
Speech Arena Elo (blind listener preference)
1,176
Elo 95% interval (±)
13
Arena rank
23
Arena rank range
20-25
Arena appearances
2,169
Voices tested
8
Price per 1M characters
47.8 USD
Released
Sept 2026
Artificial Analysis model
Gradium TTS (Sep 2026)
Artificial Analysis leaderboard
artificialanalysis.ai

Price (all sources)

Lowest list price per 1M characters
35.9 USD
Highest list price per 1M characters
57.8 USD

Frontiers

On the Speko speed and naturalness frontier
✓
On the Coval and Artificial Analysis frontier
✓

Similar TTS models