Model releases.

A mixture-of-experts model for enterprise text and image Q&A and summarization.

A 27B multimodal model with a 256K context and strong vision-language reasoning.

Generates and edits images from text with precise instruction-following.

Open-weight 1T model for long-horizon agentic coding, with 262K context and vision.

A 35B multimodal model built for top-tier vision-language reasoning tasks.

Generates video with synced audio in fewer sampling steps, using a Gemma 3 text encoder.

Generates full-arrangement music compositions with rich, layered instrumentation.

Turns the text in a flat design into editable text layers, saved as an SVG.

A 26B mixture-of-experts model built for long context and strong multimodal reasoning.

A flagship dense model with 256K context for top-tier text and image understanding.

Handles text, image, and audio input in an efficient 2B parameter multimodal model.

Compact multimodal model for text, image, and audio, with stronger reasoning than E2B.

Upscales an attached image from 2x to 16x, with optional detail and realism enhancement.

Generates detailed, prompt-faithful images using a high-quality diffusion model.

A professional image model for high-fidelity, prompt-faithful generation and editing.

Generates short video clips with native audio at a lower cost than the faster tier.

Coding model for code generation and agentic programming, with tool calling support.

A professional grade music model offering enhanced fidelity and musical expression.

Reasoning model for complex analysis, coding, and agentic tool use, with 256K context.

Converts text to expressive speech with multiple voices and support for 20 languages.

Open text-to-speech with bracket cues, multi-speaker dialogue, and voice cloning.

A frontier text model built for advanced reasoning and broad multimodal capabilities.

Handles fast on-device text and image reasoning with vision input and 256K context.

A compact 4B multimodal model with vision input and strong multilingual reasoning.

A 9B multimodal model balancing vision-language reasoning with on-device performance.

Generates images quickly, optimised for speed and creative results.

A 24B foundation model built for efficient reasoning over long text contexts.

A highly capable multimodal model built for deep reasoning with a 1M token context.

Generates music with enhanced coherence and improved audio quality.

Generates clean, scalable SVG artwork with editable paths for logos, icons, illustrations.

Generates high-fidelity SVG vector artwork for branding and production layout work.

A lightweight image generation model built for fast, high-quality results.

An advanced video model that improves motion coherence and scene quality.

Fast video generation at 480p or 720p with synchronized audio and low latency.

Generates cinematic video clips up to 1080p with native audio, lip sync, and sfx.

Generates consistent text-to-image and single-reference edits at 1K or 2K resolution.

Generates 2K and 4K images from text and up to ten reference images.

An expressive text-to-speech model with a rich emotional and tonal range.

Open full-track music generation with lyric editing, remix, cover, and voice cloning.

Studio text-to-speech model with 332 voices, emotion control, and voice cloning.

A compact 4B diffusion model for fast, high-quality image generation.

A medical multimodal model fine-tuned for healthcare and biomedical tasks.

Transcribes speech in batches, giving word-level timing for subtitles and speaker labels.

Generates images with top fidelity, strong prompt adherence, and stable multi-edits.

A fully open 32B model with open weights and data, plus long chain-of-thought reasoning.

Generates images with strong aesthetic quality and accurate prompt fidelity.

A 14B dense edge model tuned for fast on-device reasoning and tool use.

A compact 3B dense edge model for fast on-device text generation and tool use.

An 8B dense edge model for on-device chat, reasoning, and tool use.

Generates images from text quickly and cheaply, suited for low-cost production graphics.