223 AI models. One API key.

Continue an existing video with newly generated footage — describe what happens next and get 5-15 more seconds, either appended to the source or on its own. Up to 2K.

Expand a video past its original frame — the source stays put inside a wider canvas and the surrounding region is generated to match.

Reframe a video into a new aspect ratio — the source is re-cropped and the newly exposed edges are generated to match. Takes no prompt.

Speed-optimized 2.5 with synchronized native audio — up to 4K, up to 20s, with camera motion control.

Quality-optimized 2.5 with synchronized native audio in a single pass — 720p/1080p, 6-10s, with camera motion control.

Text-to-video with synchronized native audio for product, UGC, and presenter clips.

MiniMax H3 Max lip-synced video from a portrait and an audio track — mouth movements follow the soundtrack, optionally guided by a transcript. 5-14.8s of audio, up to 2K.

MiniMax H3 Max video from a start frame with scripted camera motion — 2-12 keyframes set the camera angle, height, and distance over time. Leave the prompt blank to freeze the scene and move only the camera. Up to 15s at 1080p.

Edit videos with a text instruction — change objects, styles or scenes while preserving motion, timing and audio. Source clips up to 15 seconds; output at 24 fps, up to 720p.

Faster-than-realtime MiniMax H3 Max Turbo video from text or a start/end frame, with prompt expansion. Up to 15s at 1080p.

Upscale videos toward 4K (1.5x–3x) in precise (source-faithful) or creative (detail-enhancing) mode. Source clips up to 20 seconds and 2K.

Top-tier MiniMax H3 Max video from text, a start/end frame, or reference images, videos, and audio — refer to references in the prompt as Image 1, Video 1, Audio 1, in input order. Up to 15s at 1080p.

Ideogram 4.5 — text-to-image and drift-free iterative editing with up to 4 reference images and an optional mask.

Ideogram 4.5 precise edit — change one area and keep the rest of the image pixel-intact, optionally with a mask and reference images.

Fastest 5.0 tier — quick 2K generation with up to 10 reference images.

Generate and edit images with up to 10 references — multi-reference composition, precise local edits and text rendering, natively up to 4K.

Fastest V4.1 tier — quick raster output with 10K-character prompts.

Fast GPT Image tier — everyday generation at roughly half the latency of GPT Image 2.

Most capable GPT Image tier — premium edits and campaign-grade output, with longer generation times.

Meta's agentic image model — plans with reasoning, web and image search before rendering.

Qwen-Image 3.0 Pro (GA) — flagship text-to-image and image editing with prompt-rewrite modes and thinking mode.

Latest Grok Imagine generation — sharper detail with a low/medium quality tier.

Apply curated Picsart effect presets to a photo — multi-step Magic Flow pipelines, one tap.

Style-focused raster output with 10K-character prompts.

Multi-speaker conversation on the v4 engine — a voice per line, audio tags per line.

Most expressive voice engine — audio tags direct the delivery, 90+ languages.

Eleven v4 tuned for speed — the same audio tags and languages, lower latency.

Score a video with a soundtrack written to follow what happens on screen.

Lightweight Gemini 3.8 TTS with directable delivery for faster, high-volume speech synthesis.

Latest Gemini native text-to-speech with directable delivery and inline vocal events like <laugh> and <sigh>.

Generate a multi-speaker conversation — a voice per line — in one take.

Full-length song generation with vocals from text and image prompts, powered by Google Lyria 3.5.

Text-to-music with vocals or instrumentals from a style prompt and optional lyrics, with configurable audio encoding.

Synthesize natural speech in 20 languages — pick a named voice or clone one from a reference audio.

Synthesize natural English or Chinese speech — pick a named voice or clone one from a reference audio.

Generate music with vocals or instrumental from a text prompt.

Fast, lightweight GPT-6 model for low-latency, high-volume text tasks.

OpenAI’s most capable GPT-6 model for deep reasoning and complex tasks.

Transcribe speech from audio or video, with word timings and speaker labels.

Previous-generation Sonnet — balanced reasoning, latency, and cost.

Latest Sonnet — frontier reasoning at everyday latency and cost.

Proven fast Gemini tier — low-latency multimodal text generation.

Compact GPT-4.1 tier — fast text generation with vision input.

The smallest, fastest GPT-4.1 tier for lightweight text tasks.

Multimodal GPT-4o — solid quality at low latency, no reasoning pass.

Small, cost-efficient GPT-4o tier for high-volume text tasks.

GPT-5 reasoning model for general-purpose text generation.

Fast, lightweight GPT-5 model for high-volume text tasks.