Leonardo Phoenix 1.0 — Leonardo AI's Foundational Image Model ❓📝🔢✓ → 🖼️

Leonardo Phoenix 1.0 is Leonardo AI's first foundational image generation model. It produces high-quality visuals up to roughly 5 megapixels across photorealistic and stylized looks, with strong prompt adherence — a good fit for concept art, marketing visuals, product imagery, and game assets.

What makes it distinct on Replicate is the built-in generation_mode control: you pick fast, quality, or ultra to trade render time against fidelity, instead of hand-tuning step counts.

How it compares to other image models on Replicate

Key inputs that matter

• prompt (required) — the scene description.
• generation_mode — fast, quality, or ultra; the main speed-vs-fidelity dial.
• style — apply one of Leonardo's preset visual styles.
• aspect_ratio / num_images — output shape and batch size.
• contrast — tune tonal punch.
• prompt_enhance — let the model expand a short prompt into a richer one.

When to use this model

• You want high-resolution output (up to ~5MP) without upscaling.
• You like Leonardo's preset styles and want quick, consistent results.
• You want a simple quality dial rather than tuning sampler steps.

When to reach for something else

• You need LoRA / custom fine-tunes — use SDXL's open ecosystem.
• You need the absolute fastest drafts — flux-schnell is built for speed.
• You need dependable text inside images — ideogram handles typography better.

Limitations and gotchas

• Closed model — no open weights, no self-hosted fine-tuning.
• Styles are limited to Leonardo's presets rather than arbitrary LoRAs.
• Ultra mode noticeably increases render time versus fast mode.