AI Models Roster & Live Rates

Cinematic text- and image-to-video with native audio, start and end frames. Billed per second.

Google's Veo 3.1 with synchronized sound, in 4, 6 or 8 second clips. Billed per second.

Native 30-second single takes with synchronized audio and reference images. Billed per second.

Video with native dialogue and sound, grounded in Gemini world knowledge. Billed per second.

Gemini Omni Flash 1.1 guided by up to four reference images. Billed per second.

Edit an existing clip with a plain-language instruction.

Faster, cheaper Kling 3.0 with multi-shot support. Billed per second.

Seedance 2.0 at lower latency and cost, with native audio. Billed per second.

Smooth motion and strong scene fidelity with native audio, up to 30 seconds. Billed per second.

xAI's Grok video model, 1 to 15 seconds in seven aspect ratios. Billed per second.

Fal's tuned H3: closer prompt following and better aesthetics at the same 768P. Billed per second.

The successor to Hailuo: strong motion at a low per-second price. Billed per second.

The lowest-cost Veo 3.1 tier, still with synchronized sound. Billed per second.

Transfer movement from a reference video onto any character image.

Higher-fidelity motion transfer from a reference video onto a character image.

Google's fast image model with strong prompt following and editing.

Google's latest fast image model, with web-grounded generation and editing.

Gemini 3 Pro Image: the highest-fidelity Google image model, with precise text rendering.

OpenAI's current image model: natural lighting, rich detail and precise edits.

FLUX.2 [pro]: photorealistic generation and reference-based editing.

Seedream 5.0 at an everyday price: sharp detail and editing.

Posters, thumbnails and logos with crisp, accurate typography.

Design-grade illustrations and brand-consistent graphics.

A fast, low-cost photoreal model from Alibaba Tongyi. Good for drafts and trying ideas.

FLUX.2 [klein] 9B: the compact Flux model. Quick, cheap, and it can edit your photo.

The most expressive ElevenLabs voice model. Billed per 1,000 characters.

Natural, studio-quality narration in many languages. Billed per 1,000 characters.

Fast, expressive speech with 30 voices at a low price. Billed per 1,000 characters.

Speak your script in a voice designed from a written description. Billed per 1,000 characters.

Clone a voice from a short sample and hear it speak your text.

Studio-quality songs and scores up to a minute long.

Full songs with sung lyrics. Add [verse] / [chorus] lines for vocals, or leave them out for an instrumental.

Google DeepMind's latest music model, across almost any genre.

Stereo music loops and beds, 30 seconds per generation.

Any sound effect from a description: impacts, ambience, foley, UI sounds.

Type a script and a portrait speaks it — voice included. About 6 seconds per clip.

Animate a portrait to your own audio. Billed per second of audio.

Re-sync the lips in an existing video to new audio. Billed per second of video.