Iris-3B vs Leonardo.Ai API

Iris-3B

Code and weights are Apache-2.0, and one Python package runs text-to-image, a 4-step Turbo, depth and 4x restoration from a documented CLI. It is ten days old, with no PyPI package, version tags, changelog or CI, needs an NVIDIA GPU and about 12 GB of weights, and has no server or API of its own.

Leonardo.Ai API

Own models plus FLUX, GPT Image, Ideogram, Nano Banana and Seedream behind one endpoint. No per-image price list outside the logged-in calculator.

Iris-3B

1. Check for an NVIDIA GPU with CUDA and Python 3.11 or newer before downloading. The base weights are about 12 GB, and Qwen3-VL-4B downloads on the first run
2. Exclude depth/*, upscaler/* and turbo/* in hf download unless those models are needed. Each of the first two adds about 12 GB
3. Don't pass --steps, --cfg-scale, --negative-prompt, --order or --shift with a Turbo checkpoint. The script exits with an error
4. Pass --seed to reproduce an image, and set --outdir, since files are named after the prompt and a repeat run overwrites them
5. Load only exported directories (model.safetensors and config.yaml). A training .pth is read with torch.load(weights_only=False), so never load one from an untrusted source

Leonardo.Ai API

1. Use POST /api/rest/v2/generations with a model name such as lucid-origin
2. Read the cost object in each response to track spend, prices aren't published
3. Send quality, not mode. mode was retired on 2026-05-04
4. Set a webhook callback URL on the key instead of polling long jobs
5. Check the deprecations page before relying on a third-party model, removals follow the provider