No plans, no slabs, no monthly minimums. You pay for the characters you synthesise: Feather for high-volume work, Starling for studio quality and cloning.
All 107 languages and every voice included, at the same rate. Against published list prices that is up to 50× cheaper than frontier voice models: press play below and watch the bill. Billed per character, no minimums, cancel any time.
Published list prices, normalised to the same two units, for the exact models you can hear below. The difference is compute: everyone else rents GPUs, CelestLabs runs on ordinary CPUs.
Each row is a real recording of the same line from that provider's model. The meters count up in lockstep with playback: each number is that model's price per million characters, not the cost of this clip.
FIG. 01 · eight engines, published list prices per 1M characters, each paired with the model you hear.
Published list prices per 1M characters · multiples measured against feather
This is not a loss leader, an introductory rate, or a thin reseller margin. It is what synthesis costs when the whole path runs on commodity CPUs.
The full methodology, model card, and per-machine latency table live on the research page.