Test any model from one request surface.

TypeSafe's System One decision model: send app state plus typed questions, get typed answers with calibrated probabilities in under a second.

$0.047 per 1M input tokens

Watches and listens to a video, then answers questions about it: shots, action, speech, on-screen text, lip sync.

$2.22 / $13.29 per 1M tokens

ByteDance's Seedream 5.0 Pro layer decomposition — a base image plus up to 16 transparent layers, each with its name, z-order, and bounding box.

$0.037 / layer

Alibaba's open layer-decomposition model — splits one image into 2 to 10 editable RGBA layers.

$0.028 / layer

Re-shoots one image from another camera angle — orbit, elevation and zoom — keeping the subject and scene.

$0.04 / image

ByteDance's temporally consistent depth estimation — turns any video into a flicker-free grayscale depth video.

$0.063 / second

MiniMax fast, natural multilingual speech, streamed as it is spoken; strong on Chinese.

$72.79 / 1M characters

MiniMax highest-fidelity multilingual speech for finished narration, streamed as it is spoken.

$121.32 / 1M characters

ByteDance expressive speech with sound effects, from a single script.

$0.329 / request

ElevenLabs most expressive multilingual speech model.

$103.50 / 1M characters

Google controllable multi-speaker text-to-speech.

$11.50 / 1M characters

OpenAI low-latency, steerable text-to-speech.

$13.80 / 1M characters

ElevenLabs low-latency speech for realtime agents.

$51.75 / 1M characters

OpenAI fast, low-cost speech-to-text for dictation and voice input.

$0.0000607 / second

OpenAI most accurate speech-to-text, for noisy audio, accents and mixed languages.

$0.000121 / second

Alibaba Qwen speech-to-text, strong on Chinese, its dialects and English, for dictation.

$0.0000425 / second

Suno full-song generation with vocals and instrumentation — every request returns two complete takes.

$0.152 / request

Suno's lighter, faster V6 tier for drafts and short clips — two takes per request.

$0.152 / request

Suno's experimental V6 variant with bolder, less conventional arrangements — two takes per request.

$0.152 / request

Google high-fidelity instrumental music generation.

$0.069 / second

ByteDance controllable music generation and editing.

$0.034 / second

Voyage AI general-purpose multilingual embeddings for retrieval, 1024 dimensions by default.

$0.066 per 1M input tokens

Voyage AI low-cost multilingual embeddings for retrieval, in the same vector space as Voyage 4.

$0.022 per 1M input tokens

Alibaba Qwen multilingual text embeddings, 100+ languages and strong on Chinese, with a vector length you choose.

$0.011 per 1M input tokens

BAAI open multilingual text embeddings, 1024 dimensions, for retrieval across 100+ languages.

$0.011 per 1M input tokens

OpenAI small, low-cost text embeddings for search and retrieval, with a vector length you choose.

$0.022 per 1M input tokens

OpenAI most capable text embeddings, multilingual, with a vector length you choose.

$0.144 per 1M input tokens