Speechify
simba-3.2
Measured 2026-07-03
Run simba-3.2 on Speko →One API, every model on this board.
Naturalness
1573
Win % · 95% CI
63%
57–69%
Drift
7
Synth p50
85ms
85–103 p50–p90
Re-measured 2026-09-21: two n=30 sweeps at concurrency 1 from US East, p50 82 and 87, p90 102 and 104. The pair missed the ≥90%-within-10% publish gate at 3/4 — the one failure is the co-swept deepgram:aura-2 anchor (26%), not this row, which agreed to 5.9% here and 0.0% on an earlier pair. Eight sweeps across two regions span 82–90ms, and a vendor-direct probe reads 56ms from the same region.
Robustness
0.21
Fined:
• fatal · currency — never said “four hundred fifty thousand” (49 of 49 transcripts)
→ we sent “The budget is $450K for the year.”
→ it said “The budget is four hundred and fifty k for the year.”
• fatal · decimals — dropped the minus sign (7 of 7 transcripts)
→ we sent “Growth slowed to -0.5% this month.”
→ it said “Growth slowed to 0.5% this month.”
• fatal · large numbers — never said “one million one” (49 of 49 transcripts)
→ we sent “We logged 1,000,001 events overnight.”
→ it said “We logged a thousand thousand one events overnight.”
+ 28 more fined
Cost / 1M chars
$10.0
Vendor list (Starter tier); $6–8 per 1M at Pro/Scale volume.
Share this result
Embed the live badge
[](https://benchmarks.speko.ai/tts/speechify-simba-3-2)
<a href="https://benchmarks.speko.ai/tts/speechify-simba-3-2"><img src="https://benchmarks.speko.ai/badge/tts/speechify-simba-3-2.svg" alt="Speko tts rank"></a>
https://benchmarks.speko.ai/tts/speechify-simba-3-2