Midjourney
AI image generator that creates high-quality art and photos from text prompts via web and Discord
AI image generator known for strong prompt adherence, readable text, and native 4K output
Reve (formerly codenamed Halfmoon) is an AI image generator that competes on precision rather than pure aesthetics — it topped an independent text-to-image leaderboard at launch, beating Midjourney, Flux, and Ideogram on prompt adherence. The current Reve 2.0 model renders natively at 4K and excels at in-image text. A free tier gives ~20 generations per day; paid plans are Lite at $7.99/mo and Pro at $19.99/mo, based on a monthly "energy" allowance. Best for accurate, text-heavy, editable images.
Reve is an AI image generator that competes on precision rather than pure artistic flourish. Launched in preview under the codename “Halfmoon,” it topped an independent text-to-image leaderboard on release — ahead of Midjourney, Flux, and Ideogram — on the strength of two things: prompt adherence and text rendering. In a prompt like “a warrior holding a sword next to a wizard holding a staff,” Reve keeps each subject’s attributes bound correctly, where many models blur them together. A dedicated typography pass also makes embedded text legible, closing a gap that long frustrated Midjourney users.
Under the hood, the current Reve 2.0 model takes a layout-first approach: it plans a structured intermediate representation — object positions, sizes, per-object descriptions, and color — before rendering. That design enables native 4K (4096x4096) output and lossless iterative editing, so you can refine an image repeatedly without the generational quality decay that plagues regenerate-and-hope workflows. Reve pairs this with a direct-manipulation editing suite, multi-reference image support, and an “agent-native” API in beta. Pricing dropped the old fixed-credit packs in favor of a monthly “energy” allowance across free, Lite, and Pro subscriptions.
Reve is for people who need images to be correct, not just pretty — where the prompt has specific requirements, the text has to be readable, and the output may go through several edits. Its free tier is enough to evaluate the model, but the daily energy refresh is too tight for steady production, so regular users move to Lite or Pro.
Starting price: $7.99/mo · Free tier: yes · Model: freemium
Price history tracked from June 2026
| Plan | Price | Includes |
|---|---|---|
| Free | Free | Introductory energy allowance plus daily refresh (~20 generations/day) · Full model access and community feed · Basic storage |
| Lite | $7.99/mo | About 5x the energy of the free tier · 5x storage and private images · Cancel anytime |
| Pro | $19.99/mo | About 100x the energy of the free tier · 100x storage and priority processing · 250 video energy per month (video generation option) |
| API (beta) | ~$0.0067/image | Roughly $0.0067 per image via the developer console · Create, edit, and remix operations (5 credits each) · Extra cost for upscaling, background removal, and longer generations |
| Pros | Cons |
|---|---|
| Leading prompt precision — binds attributes correctly where rivals blend them | Aesthetic polish still trails Midjourney, and it's weaker on photorealistic faces and complex hands |
| Readable embedded text, an area Midjourney historically struggled with | Free tier's daily refresh is too tight for production workloads |
| Native 4K output and a free tier of about 20 generations per day | Newer entrant with a smaller community, no mobile app, and an API still in beta |
| Lossless iterative editing avoids the quality decay of repeated regeneration |
AI image generator that creates high-quality art and photos from text prompts via web and Discord
AI image generator specializing in accurate text rendering and layout-aware design output
API-first AI image generator from Black Forest Labs with per-image pay-as-you-go pricing
AI image and vector generator with custom styles, mockups, and design-ready output
AI image generator with custom model training, canvas editor, and token-based generation
Yes. Reve has a free tier that grants an introductory energy allowance plus a daily refresh, worth roughly 20 generations per day, with full model access. Heavier use needs the Lite plan at $7.99 per month or Pro at $19.99, which multiply the energy allowance and add private images and priority processing.
Reve's headline strength is prompt adherence — it binds attributes correctly in multi-subject prompts where rivals mix them up — plus readable in-image text. It generates natively at 4K resolution and plans a structured layout before rendering, which enables lossless iterative editing without quality decay across edits.
Reve offers a free tier, a Lite plan at $7.99 per month, and a Pro plan at $19.99 per month. Higher tiers multiply the monthly energy allowance rather than adding fixed credits — Lite is about five times the free allowance and Pro about one hundred times, with Pro also including monthly video energy.
It depends on the goal. Reve wins on prompt precision, text rendering, native 4K output, and price, and it topped an independent text-to-image leaderboard at launch. Midjourney still leads on raw artistic polish and photorealistic faces, so Reve suits accuracy and Midjourney suits aesthetics.
Yes, Reve offers an API in beta through its developer console, priced around $0.0067 per image with create, edit, and remix operations. Post-processing such as upscaling and background removal costs extra. Exact per-image rates appear inside the authenticated console rather than on the public site.
They are the same product. The older reve.art domain now permanently redirects to reve.com, which is the canonical site, with the app hosted at app.reve.com. Several unaffiliated lookalike domains publish fabricated pricing, so reve.com is the address to trust.