Changelog

Follow up on the latest improvements and updates.

RSS

Flash V2 is an update to Recraft V4.1 Flash, our fastest model. Compared with the first version, it's better with text in images (signage, packaging, posters with headlines), holds fine details more reliably, and makes photo-like images look more like photographs: portraits, product shots, and interiors. It's faster too, at 1.1 seconds from prompt to image, and the price per image stays the same.
If you already use Flash, there's nothing to change: Flash V2 takes its place in the model picker. On the API, use the new model id recraftv4_1_flash_v2.
Four MiniMax H3 models are now available in Recraft Studio.
MiniMax H3 Image and MiniMax H3 Image Highspeed generate and edit images at 2K or 4K. Highspeed is the faster version.
MiniMax H3 Highspeed and MiniMax H3 Lite are video models that generate clips of 4 to 15 seconds at 480p or 768p, with sound. Highspeed balances quality and speed, and Lite is the lower-cost option for drafts.
Find all four in the model picker under MiniMax. The image models are available on plans that include external models, and the video models on plans that include video and external models.
Vectors in Recraft Studio now come out cleaner, whether you generate a vector from scratch or vectorize your own image with the new Vectorizer Pro.
Here’s what you’ll see:
Curves stay smooth
instead of breaking into angles.
Straight lines stay straight.
Corners stay sharp
instead of rounding off.
It’s available to everyone in Studio, on every plan. Select an image and click Vectorize Pro, or run an image you vectorized before through it again and compare.
Gemini Omni Flash 1.2, Google's newest video model, is now in Recraft Studio.
Compared to Gemini Omni Flash 1.1, it keeps characters, products, and scenes more consistent from the first frame to the last, follows camera directions with fewer unwanted cuts, and produces more natural dialogue with clearer lip-sync, including scenes with several speakers.
Clips run up to 10 seconds, with native audio. Find it in the model picker with the other video models. Available on plans that include video and external models.
Vary gives you four new versions of any image in Recraft Studio in one click. You don’t write a prompt. Vary writes the variation instruction for the editing model behind the scenes.
Pick how far it goes:
  • Vary direction
    keeps the core idea and tries new compositions and ideas.
  • Vary details
    keeps the composition, subject, and style, and changes only the small details.
You’ll find Vary in the Action panel, in the right-click menu on any canvas image, and in the actions on any image in chat. Each run puts four versions next to the original, ready to vary again, edit, or export.
A few places it helps:
  • Logo exploration:
    Vary direction finds new routes from a first concept, and Vary details polishes the one you pick.
  • Near-final images:
    fix small details without rewriting the prompt or risking the rest of the image.
  • Series:
    make siblings of one image for a set, a campaign, or a product line.
Vary is available on paid plans.
API Portal is the new dashboard for the Recraft API, at recraft.ai/app/api. It shows where your API units go, what happened with each request, and who on your team has access.
Here’s what’s inside:
  • Usage:
    API units spent over time, broken down by model
  • Request log:
    every request, newest first, with CSV export
  • API requests:
    4xx and 5xx errors, p50 and p95 latency, and your top endpoints
  • Workspace:
    invite teammates as admin, billing admin, or member, so everyone sees the same usage and requests
  • API tokens:
    create and revoke tokens that share one account balance
  • Balance:
    check your balance and top up API units
  • Code examples:
    requests in several programming languages
Your API tokens and balance have moved here from the API tab in your profile. Only requests made with API tokens show up in the portal. Studio and MCP activity doesn’t.
GPT Image 2.6, OpenAI's newest image model, is now in Recraft Studio in two versions.
Flare
is built for fast drafts.
Sunburst
keeps edits consistent, so the details you didn't ask to change stay untouched.
Generate at 1K or 2K, then edit, upscale, or vectorize right on your project canvas. Turn on model comparison to see it next to any other model on the same prompt.
Sunburst is available on the Free plan with default settings. Flare and all quality levels come with paid plans.
Start with your business, not a prompt.
Recraft Logo Maker asks a few questions about your brand and turns the answers into logo directions you can actually use. No prompt writing, no design experience needed.
Here’s how it works:
  • Tell us your brand name, what you do, the type of logo you want, and the traits it should carry
  • Pick a style curated by Recraft designers, or upload your own references
  • Review four different directions, each generated with the model best suited to it
  • Edit your favorite, change its colors, or ask for variations
  • Preview it on packaging, apparel, and storefronts at no extra generation cost
  • Export it as a vector file (SVG), ready for print and digital use
Logos made on the Free plan are public and for personal use. Paid plans include full ownership and commercial rights.
Create your logo at recraft.ai/logo.
Finally, zero lag in your workflow! Recraft V4.1 Flash generates images in 1.3 seconds end to end, at a fraction of the cost of standard models.
And when you're ready to publish, use the new "Refine" button to take the image to high-definition standard.
See what you can do with Recraft V4.1 Flash.
OpenAI's fastest image generation and editing model is available in Recraft Studio now.
It delivers higher-quality image generation and editing than its predecessor GTP model at 50% lower latency, which means faster iteration without accepting worse output as the cost of moving quickly.
It's built for the kind of work where volume and quality both matter: creator content, product visuals, visual search, rapid prototyping, and high-volume generation pipelines where outputs need to stay consistent across the run.
Try it out in Recraft Studio today.
Load More
→