Leonardo AI drops Phoenix to challenge Ideogram

Prompt adherence has been the biggest pain point in AI image generation since day one. You ask for a specific painterly style and legible English text, and you usually end up with generic digital gloss and unreadable alien gibberish.

On Monday, Leonardo AI rolled out its new flagship model, Leonardo Phoenix, alongside a complete overhaul of its web platform. The company did not hold back in its announcement, claiming Phoenix is among the best in the industry at faithfully executing complex user prompts.

Sharp on details, but photorealism still lags

Phoenix targets four core upgrades: stricter prompt following, legible in-image text, higher image quality, and granular creative control. Because of the brand new architecture, a few existing features like image guidance, elements, and photoreal modes are temporarily unsupported. Token costs are discounted for the launch week, reverting to regular pricing next Monday.

Head-to-head testing against Ideogram reveals an interesting dynamic. In an art movement test combining a Van Gogh painting style with custom sign text, Leonardo captured the iconic brushwork clearly and nailed the lettering, while Ideogram produced a safer, generic scene.

When pushed with complex spatial positioning, both models handled object placement accurately. Leonardo won on precision, generating proper human hands and structured objects, but Ideogram took the aesthetic crown with a more cohesive overall composition. In close-up photorealistic portraiture, Ideogram still held onto its lead with cleaner facial realism.

Beyond generation: the full workstation approach

Where Leonardo really pulls away is not raw pixels, but the operational suite wrapped around them.

The standout addition is Edit with AI, which allows creators to adjust existing generations using conversational prompts like making a coat red or shifting a scene into a 1970s vintage aesthetic on the first try. It behaves much more dynamically than standard image-to-image workflows.

Leonardo also caps 16:9 images at 1376 by 768 pixels compared to Ideogram's 1280 by 720, and adds convenient social media aspect ratio presets. Factor in built-in background removal, upscaling, real-time generation, canvas inpainting, and custom user model training, and the gap in utility becomes obvious. That utility comes at a premium, with Leonardo's Artisan Unlimited plan running at 30 dollars a month against Ideogram's 20 dollar tier.

Why it matters

We think the AI image race is moving away from slot-machine prompting and into direct editorial workflow. Generating a breathtaking still is no longer rare; the real friction for working creators has always been adjusting specific flaws without destroying the rest of the canvas.

Leonardo played this move smartly. Instead of betting everything purely on raw benchmark output, it built an all-in-one studio where prompt expansion, text instruction editing, and post-processing live in a single tab. For designers, marketers, and power creators who need fast, usable assets rather than lucky one-offs, it is definitely worth testing right now.

The trade-off is clear. Dropping legacy features during the architectural switch might frustrate existing power users, and raw photorealism still looks like Ideogram and Midjourney territory. But if Leonardo keeps polishing its instruction-based editing, it could turn into the standard interface for how people actually work with synthetic media.