Introduction

Recraft has shipped Recraft V4.1 Flash V2, a closed text-to-image model that replaces the original V4.1 Flash in Recraft Studio's model picker and on the Recraft API. The pitch is simple and very creator-friendly: roughly 1.1 seconds from prompt to image, cleaner text inside the picture, steadier fine detail, and the same $0.007 per image price as before. You can run it today in Recraft Studio or through the API with the new model id recraftv4_1_flash_v2.

When we covered the first Flash back in September, it was clocking about 1.3 seconds. Flash V2 shaves a little more off that, but the bigger story is quality: Recraft says it now handles the jobs that fast models usually fumble.

What Changed in Flash V2

Better text in images

Recraft's Nov 13 changelog calls out signage, packaging, and posters with headlines as the places where Flash V2 beats the first version. If you mock up a lot of product labels or event flyers, that is the upgrade to test first.

Finer detail and more photographic photos

The changelog also says Flash V2 holds fine details more reliably and makes photo-style images look more like real photographs, specifically portraits, product shots, and interiors.

Speed, independently checked

Recraft says @DesignArena independently confirmed that nothing on its board generates faster, and that Flash V2 sits on DesignArena's Pareto frontier for preference versus speed.

Specs at a Glance

  • Model id: recraftv4_1_flash_v2
  • Speed: about 1.1 s end to end per image (Recraft model docs)
  • Price: $0.007 per image, which is 7 API units ($1 = 1,000 units)
  • Output: 1K raster, WebP or PNG
  • Limits: raster only, no vector output and no styles
  • Access: closed model, Recraft Studio plus the pay-as-you-go Recraft API
  • Migration: in Studio, Flash V2 takes the old Flash's place in the picker, so you don't need to change anything

Recraft's docs list photography, portraits, atmosphere, complex scenes, mixed media, and logo concepts as its strongest use cases, and say it's built for high-volume generation and rapid iteration.

Try It on the API

Create a Recraft API token (your account needs a positive API unit balance), then send a request. Recraft recommends "response_format": "multipart" with Flash V2 so the image bytes come back in the same response, with no second download:

export RECRAFT_API_TOKEN=your_token

curl https://external.api.recraft.ai/v1/images/generations \
  -H "Authorization: Bearer $RECRAFT_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "a neon ramen shop sign that reads OPEN LATE, rainy night street, photographic",
    "model": "recraftv4_1_flash_v2",
    "response_format": "multipart"
  }' \
  --output response.multipart

Prefer the OpenAI Python SDK? Point it at Recraft's base URL. The SDK can't read multipart responses, so this version returns a URL instead:

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://external.api.recraft.ai/v1",
    api_key=os.environ["RECRAFT_API_TOKEN"],
)
response = client.images.generate(
    prompt="minimalist coffee bag packaging with the headline MORNING ORBIT, studio product shot",
    model="recraftv4_1_flash_v2",
)
print(response.data[0].url)

A Starter Prompt for Text Tests

Paste this in to see how Flash V2 handles lettering:

A vintage cinema marquee at dusk with the glowing headline MIDNIGHT DOUBLE FEATURE, warm bulbs, wet pavement reflections, photographic, sharp lettering

Original Source

Recraft's changelog entry: https://recraft.canny.io/changelog

Model docs: https://www.recraft.ai/docs/api-reference/models/recraft-v4-1-flash-v2

Conclusion

Flash V2 is the kind of quiet upgrade that adds up quickly: the same tiny price, a little more speed, and fewer garbled headlines in your mockups. If you're iterating on posters, packaging, or product shots, a $0.007 test batch costs pennies and tells you a lot. Go make something fast and legible!

—Aurelia ♡