# Reve Image Guide to Halfmoon, Model History and Prompting > Published 2026-04-05, updated 2026-09-05 · https://www.promptzone.com/divya_watanabe/halfmoon-reve-advanced-ai-image-tool-41e4 --- title: "Reve Image Guide to Halfmoon, Model History and Prompting" published: true description: Understand the Halfmoon name behind Reve Image, distinguish its preview from later models, and test a composition and typography brief. tags: ai,imagegeneration --- Reve Image is the hosted image generator Reve identified as Halfmoon on March 24, 2025. The current web app presents Reve 2.1. [Announcement data][announcement-data] [Current model][model] ## What are the key facts about Reve Image and Halfmoon? | Field | Verified information | | --- | --- | | Developer | Reve AI, Inc. [Company page][about] | | Released | Public Halfmoon-to-Reve announcement on March 24, 2025. [Official announcement][announcement] | | Type | Hosted text-to-image generation, introduced publicly as Reve Image. [Announcement data][announcement-data] | | Size or parameters | No parameter count in the Halfmoon announcement. [Announcement data][announcement-data] | | License and access | Proprietary hosted access; no open model weights. [Service terms][terms] [Model documentation][model] | | Where it runs | Reve's web service; the current app presents Reve 2.1, without a documented Halfmoon selector. [Model documentation][model] | Use the March 2025 announcement to identify Halfmoon, and the current product documentation to choose an access path. Record the model label shown when you generate a new image. [Announcement data][announcement-data] [Model page][model] ## How does Reve describe composition and text generation? Reve's retrospective model description says 1.0 was trained with structures describing composition, relationships, style, and text. That provides a concrete way to understand the product's emphasis on visual organization. It does not publish a task-specific success rate or establish that every requested arrangement will be correct. [Model documentation][model] A practical composition brief should state what belongs in the image and how its parts relate. For an independent bookstore event, specify the main object, headline, supporting line, and space around each element. Decide which component should attract attention first before requesting a stylistic treatment. Try this original exercise: an illustrated open book centered on a cream poster, a large headline above it, and a short location line below. Judge whether the hierarchy is clear before evaluating the illustration's decorative details. Use this as a brief for a current generation session. For the service's later high-resolution release, the sibling [Reve 1.5 guide](/santiago_eriksson/reve-15-released-enhanced-ai-art-with-stable-diffusion-22ie) explains what that launch documented. Keeping these releases separate helps you choose which capabilities need to be verified for a particular assignment. ## Can you still access the original Halfmoon model? The Halfmoon announcement identifies Reve Image but supplies no architecture specification or parameter count. [Announcement data][announcement-data] The current model page presents Reve 2.1 and discusses Reve 1.0 retrospectively. It does not document a selector for the March 2025 preview. [Model documentation][model] Reve's API closed on August 14, 2026. Its help center documents the current web application for image creation. [API status][api-status] [Getting started][start] For a historical comparison, keep the generated image, original prompt, and known model identity together. If one of those is missing, label the evidence accordingly. A contemporary reconstruction can illustrate a creative idea, but it cannot establish how the original service behaved under unknown settings. ## How do you test a composition brief in Reve today? Use Reve's current web app for a new creative session. The documented service offers no open Halfmoon weight download. [Getting started][start] [Service terms][terms] The public API is closed. [API status][api-status] Begin by deciding whether you are researching the model's history or making a new asset. For historical work, use the dated vendor announcement and retained output records. For new artwork, continue in the current app and record the model label it presents. The current help guide documents entering an idea directly in chat and sending the request. Use plain language to define the intended layout, then review the image before continuing the conversation. [Chat instructions][chat] For the bookstore exercise, write an original prompt such as: “A cream literary-event poster. An open blue book occupies the center. Large navy text above the book reads 'READ TOGETHER'. A smaller line below reads 'RIVER STREET BOOKS'. Leave generous outer margins.” Inspect the result in a deliberate order. First read the generated text character by character. Then check whether the headline is dominant, whether the book is visually distinct from the lettering, and whether the margins provide room for the intended placement. Keep a short list of requirements rather than an overall impression. For this example, the requirements are the exact headline, the exact supporting line, an open blue book, a cream background, and a clear vertical hierarchy. A result can pass some requirements while failing others. If the composition fails, simplify the request. Try the book and headline without the supporting line, then evaluate whether the central relationship is clearer. If the words fail, compare shorter lettering or reserve a clean area for manually typeset text in your final design. These are suggested production choices rather than guaranteed model fixes. When refining through the current app, use its documented image-selection control to bring the chosen result into chat before giving another instruction. Keep the accepted elements in your review checklist. A revision needs to preserve them as well as address the requested change. [Chat instructions][chat] Inspect the poster at its final display size. Save the approved image with its prompt and model label, and record any manually replaced typography. ## How does Reve compare with Recraft for design outputs? Recraft V4 is a concrete alternative for a design brief that requires editable SVG output: its official documentation lists vector model variants alongside raster generation. Reve's retrospective description of its 1.0 model focuses on structured image generation, while the present Reve service offers its own visual workflow. Compare the output artifact you need before comparing stylistic preferences. [Recraft V4 documentation][recraft] [Model documentation][model] If you are considering a local workflow, the [ComfyUI pillar](/tomas_novak/comfyui-2026-the-complete-guide-to-power-user-ai-image-generation-1g17) provides the broader context. Establish whether local control, a hosted editor, or editable vectors are required, then test the same creative brief in the appropriate tool. ## What should you know about the Halfmoon name? ### Is Halfmoon the same as Reve Image? Reve publicly identified Halfmoon as Reve Image on March 24, 2025. That announcement establishes the preview's identity. [Official announcement][announcement] [Announcement data][announcement-data] ### Where can I download Halfmoon weights? Reve documents hosted access and provides no open-weight download for Halfmoon. Its terms reserve the underlying service models to Reve. [Getting started][start] [Service terms][terms] ### Can I reproduce a Halfmoon image in the current Reve app? Reve's current model page presents Reve 2.1 and does not document a Halfmoon selector. Check the available version before attempting to reproduce an original preview output. [Model documentation][model] ### What did Reve publish about the Halfmoon model's size? Reve's Halfmoon announcement does not publish a parameter count. The current model page discusses later releases without giving an absolute count for the preview. [Announcement data][announcement-data] [Model page][model] ## Sources - [Official announcement][announcement] - [Announcement data][announcement-data] - [Model documentation][model] - [Company page][about] - [Service terms][terms] - [API status][api-status] - [Getting started][start] - [Chat instructions][chat] - [Recraft V4 documentation][recraft] [announcement]: https://x.com/reve/status/1904211082870456824 [announcement-data]: https://cdn.syndication.twimg.com/tweet-result?id=1904211082870456824&token=0 [model]: https://app.reve.com/model [about]: https://app.reve.com/about [terms]: https://app.reve.com/terms [api-status]: https://help.reve.com/hc/en-us/articles/46837930295316-Reve-API [start]: https://help.reve.com/hc/en-us/articles/46809604700052-Getting-started-with-Reve [chat]: https://help.reve.com/hc/en-us/articles/46776083658132-Creating-images-in-chat [recraft]: https://www.recraft.ai/docs/api-reference/models/recraft-v4 ## Related guides on PromptZone - [Best SDXL Models in 2026](/tara_suzuki/best-sdxl-models-in-2026-realistic-anime-and-all-purpose-checkpoints-116) - [ComfyUI 2026: The Complete Guide](/tomas_novak/comfyui-2026-the-complete-guide-to-power-user-ai-image-generation-1g17) - [AI Model Releases Timeline](/ai-model-releases)