The newest voice synthesis model from Speechify topped the quality rankings for voice control in a recent evaluation of text-to-speech tools, finishing statistically level with the equivalent model from HeyGen. In the results published by Artificial Analysis, Simba 3.5 surpassed the performance of Qwen-Audio-3.1-TTS-Plus to secure the top spot in the Controlled Voice Arena.
Simba 3.5 also recorded the lowest cost on the analyst’s price chart at $6.58 per 1 million characters, significantly undercutting the $40 price point of Eleven’s v4 Turbo. It generate 139.6 characters per second, beating Eleven’s speed of 105.3 characters per second. While it led in control, the model ranked fifth in provider voice and seventh in pronunciation robustness. The model is currently accessible through the OpenRouter API and supports 48 languages.
