Raphael AI

Raphael AI debuted on Indie Hackers on April 14, 2025 from solo developer yukiji0701. It started as a one-person Flux.1-Dev pipeline running on an idle A100 with one clear mission: build a text-to-image service that is genuinely free, unlimited, no-login, and watermark-free. The name honors the Renaissance painter Raphael.

Core capabilities center on a smart multi-model router. The free tier (Raphael Basic 2.0) is unlimited but goes through the slow queue, capped at 1024px and bearing the Raphael watermark — fine for personal experiments. After login, 10 fast-queue credits per day unlock Fast Mode (about 10 images). Paid Premium at $10 per month (billed annually) unlocks the Pro model (up to 1K resolution), 2,000 fast-queue credits per month, watermark removal, and a commercial license; Ultimate at $20 per month (billed annually) adds the Max (1.5K) and Ultra (2K) models plus 5,000 fast-queue credits. The supported model pool keeps growing: as of 2026 it spans Raphael's own models, GPT Image 2, Nano Banana 2, Seedream 5, Qwen-Image, Flux 2, and Z-Image.

Use cases include everyday inspiration sketches, e-commerce product draft imagery, blog post illustrations, social-media mockups, designer concept drafts, and teaching illustrations. It is generally not the right pick for final commercial deliverables — the free tier carries a watermark at 1024px, and commercial use requires Premium or above. All uploaded images and generation prompts are automatically deleted within 10 minutes, for privacy. Chinese-prompt support is a distinctive selling point — most English-first tools handle Chinese prompts poorly. Public data points to roughly 3 million users and about 1,530 images generated per minute (as of 2026).