What makes Sunburst different from Flare?

Sunburst is slower but better at rotation, turntables and consistent fine detail; Flare is faster and better at big motion and cinemagraph loops.

What formats can it output?

png, jpeg, webp and native animated gif, with optional transparent backgrounds.

Does it render text inside the image?

Yes — legible in-image text in detailed stills up to 3840x2160.

GPT Image 2.6 Sunburst

Create looping 360-degree product turntable GIFs from a photo.

Playground

API Pricing

~36.20s

Example output

Try an example4

GPT Image 2.6 Sunburst (Text-to-Image & Animated GIF)

What is GPT Image 2.6 Sunburst?

GPT Image 2.6 Sunburst is OpenAI's detail-first image model on Segmind. One endpoint handles text-to-image, reference-guided editing and native animated GIF output. Its signature capability is the animated GIF turntable: give it one clean product photo and it returns a seamless 360-degree spin whose fine detail — patterns, materials and proportions — stays consistent from frame to frame, with no 3D model and no video pipeline. Sunburst is the precision route in the GPT Image 2.6 family: it takes longer per request than the Flare route and spends that time holding detail steady as the subject turns.

Key Features

Native animated GIF, 2 to 16 frames, with frame-duration control and a seamless loop.
Frame-to-frame consistency: pattern, colour, geometry and lighting hold as the subject rotates.
Image-to-GIF: animate a reference photo while keeping the object recognisably the same.
Detailed stills with legible in-image text, at sizes up to 3840x2160.
Targeted edits with an optional alpha-channel mask, and transparent backgrounds with png, webp or gif.

Best Use Cases

Sunburst is built for product spins: rotate ceramics, watches, packaging or collectibles for a store listing, a marketplace preview or an email that loops in place. It also suits design reviews and concept art you want to inspect from every angle, and detail-heavy stills — packaging, posters and editorial illustration with dense, readable text. In testing, a single carved-figurine photo became a clean 9-frame 960x960 turntable whose scales, gold edging and proportions matched across every frame.

Prompt Tips and Output Quality

Start from one clean subject on a plain backdrop with even light. State the motion exactly — "one full revolution around the vertical axis, fixed camera, no zoom" — and name what must stay constant: pattern, colour, proportions, lighting. Use 960x960 with 9 frames for crisp turntables, or Auto with up to 16 frames for a smoother spin. Keep quality high for finals; if a large frame size combined with many frames is rejected, lower one.

How to Use GPT Image 2.6 Sunburst

GPT Image 2.6 Sunburst is the precision route in the GPT Image 2.6 family. Reach for it when every frame of an animation must show the same object with the same detail — product turntables, slow rotations and close-up loops — and for detailed stills and precise edits. It is slower than the Flare route, so expect longer waits on high settings.

The most useful workflow is image-to-GIF. Put one clean product photo in image_urls, set output_format to gif, and describe a turntable: "one full 360-degree revolution around the vertical axis, fixed camera, no zoom." Then name what must stay constant — pattern, colour, proportions and lighting — and ask for a seamless loop. size sets one frame, so square frames keep a spin crisp. Say "no text, no logos" explicitly when you want a clean object.

For detailed stills, switch output_format to png, raise size to 1536x1024 or 2048x2048, and keep quality high; the model renders legible in-image text well. For cut-outs, set background to transparent with png, webp or gif. To repaint one region, pass an alpha-channel mask_image_url that matches your first reference.

Recommended Settings

Product turntable (crisp): reference in image_urls, output_format gif, size 960x960, frame_count 9, frame_duration_ms 200, quality high, complexity high. This is the tested default example, and it produced a clean, consistent 360-degree spin.
Smoother spin: size auto, frame_count 16, frame_duration_ms 100-150.
Slow showcase: keep 9 frames and raise frame_duration_ms to 300 or more.
Faster drafts: complexity low and quality low, then switch both back up for the final render.

If a large frame size combined with many frames is rejected, lower frame_count or the frame size.

Examples to try

Each card below ran end to end at the settings shown - click one to load its prompt and settings into the form. The first animates one uploaded photo; the rest generate from a prompt alone, so leave the image field empty.

Reference-guided image-to-GIF turntable (one reference photo). Prompt: "Animate this exact carved wooden dragon figurine as a smooth 360-degree turntable rotation on the same seamless pale-grey backdrop. The dragon turns at a steady speed around its vertical axis, one full revolution across the loop, so the horns, whiskers, clawed feet gripping the pearl and the curling tail pass in and out of view. Keep the rows of overlapping carved scales, the jade-green paint with gold-leaf edging along the spine ridge and horns, the deep-red mouth, the proportions and the even studio lighting consistent in every frame; fixed camera, no zoom. The last frame flows back into the first. No text, no logos, no labels." One clean product photo in image_urls becomes a seamless 960x960 loop where the carved scales, gold edging and held pearl read identically in every frame. Tip: name what must stay constant, then ask for a seamless loop.
Legible in-image text. Prompt: "A lit cinema marquee on a brick theater facade at night. The marquee's white illuminated board shows black changeable-letter text: 'THE LARKSPUR' in tall condensed capitals across the top, and 'NOW SHOWING TONIGHT' on the line below, framed by rows of round marquee bulbs. The lettering is evenly spaced, high-contrast and sharp. No other text, no brand names, no logos." A single 1536x1024 PNG still renders both lines crisp and correctly spelled. Tip: put the exact wording in quotes and add "no other text" to keep the frame clean.
Transparent-background cut-out. Prompt: "A neat stack of five French macarons in pastel colors - pistachio green, rose pink, lavender, lemon yellow and pale blue - each with a smooth domed shell, a ruffled 'foot' and a thick cream filling, centered with generous padding. Clean sharp alpha edges, fully transparent background, no surface, no backdrop, no shadow, no checkerboard, no text, no logos." With background transparent and output_format png the stack lands on clean alpha edges you can drop onto any layout. Tip: ask for "no backdrop, no shadow" so nothing bleeds into the cut-out.
Native text-to-GIF turntable. Prompt: "A vintage twin-lens reflex camera stands on a seamless light-grey backdrop and rotates a smooth, steady 360-degree turntable around its vertical axis, one full revolution across the loop. It is a boxy black-and-chrome body with two stacked round glass lenses, a knurled focus knob, a brushed-metal winding crank and textured black leatherette panels. Keep the lens rings, engraved dial markings, rivets and leather grain identical in every frame; even studio lighting, fixed camera, no zoom. The last frame flows back into the first. No brand names, no logos, no text." Twelve frames complete a full revolution from a prompt alone - no photo - and the dials, rivets and leather grain hold across the loop. Tip: keep GIF frames square or 3:2 and name every detail that must stay put.

Also built for: "repaint only one region with an alpha mask_image_url", "a 2048x2048 product still with a readable spec sheet", or "a 16-frame transparent-background looping sticker". Treat these as starting points, not tested runs.

GPT Image 2.6 Sunburst examples

Requests run on GPT Image 2.6 Sunburst in the playground: what each was given, what came back, and the API call that makes it.

How to animate a figurine photo into a 360-degree turntable loop with GPT Image 2.6 Sunburst

Prompt

Animate this exact carved wooden dragon figurine as a smooth 360-degree turntable rotation on the same seamless pale-grey backdrop. The dragon turns at a steady speed around its vertical axis, one full revolution across the loop, so the horns, whiskers, clawed feet gripping the pearl and the curling tail pass in and out of view. Keep the rows of overlapping carved scales, the jade-green paint with gold-leaf edging along the spine ridge and horns, the deep-red mouth, the proportions and the even studio lighting consistent in every frame; fixed camera, no zoom. The last frame flows back into the first. No text, no logos, no labels.

Input image

Settings

size: 960x960
quality: high
complexity: high
background: opaque
moderation: auto
output_format: gif
frame_count: 9
frame_duration_ms: 200

Animate this exact carved wooden dragon figurine as a smooth 360-degree turntable rotation on the same seamless pale-grey backdrop. The dragon turns at a stead… — Output from GPT Image 2.6 Sunburst

API request

# pip install "segmind>=1.1.0"
# export SEGMIND_API_KEY="YOUR_API_KEY"
import segmind

result = segmind.run(
    "gpt-image-2.6-sunburst",
    prompt="Animate this exact carved wooden dragon figurine as a smooth 360-degree turntable rotation on the same seamless pale-grey backdrop. The dragon turns at a steady speed around its vertical axis, one full revolution across the loop, so the horns, whiskers, clawed feet gripping the pearl and the curling tail pass in and out of view. Keep the rows of overlapping carved scales, the jade-green paint with gold-leaf edging along the spine ridge and horns, the deep-red mouth, the proportions and the even studio lighting consistent in every frame; fixed camera, no zoom. The last frame flows back into the first. No text, no logos, no labels.",
    image_urls=[
        "https://segmind-resources.s3.amazonaws.com/input/gpt-image-2.6-sunburst-input-dragon-dcf844f7.jpg",
    ],
    size="960x960",
    quality="high",
    complexity="high",
    background="opaque",
    moderation="auto",
    output_format="gif",
    frame_count=9,
    frame_duration_ms=200,
    output_compression=100,
)
print(result.get("output"))

# 1. Submit the request; the reply carries its request_id
REQUEST_ID=$(curl -s -X POST "https://api.segmind.com/v2/gpt-image-2.6-sunburst" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "Animate this exact carved wooden dragon figurine as a smooth 360-degree turntable rotation on the same seamless pale-grey backdrop. The dragon turns at a steady speed around its vertical axis, one full revolution across the loop, so the horns, whiskers, clawed feet gripping the pearl and the curling tail pass in and out of view. Keep the rows of overlapping carved scales, the jade-green paint with gold-leaf edging along the spine ridge and horns, the deep-red mouth, the proportions and the even studio lighting consistent in every frame; fixed camera, no zoom. The last frame flows back into the first. No text, no logos, no labels.",
  "image_urls": [
    "https://segmind-resources.s3.amazonaws.com/input/gpt-image-2.6-sunburst-input-dragon-dcf844f7.jpg"
  ],
  "size": "960x960",
  "quality": "high",
  "complexity": "high",
  "background": "opaque",
  "moderation": "auto",
  "output_format": "gif",
  "frame_count": 9,
  "frame_duration_ms": 200,
  "output_compression": 100
}' | jq -r .request_id)

# 2. Check its status until it is COMPLETED
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID/status" \
  -H "x-api-key: YOUR_API_KEY"

# 3. Fetch the result
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID" \
  -H "x-api-key: YOUR_API_KEY"

How to render custom text on a cinema marquee sign with GPT Image 2.6 Sunburst

Prompt

A lit cinema marquee on a brick theater facade at night. The marquee's white illuminated board shows black changeable-letter text: 'THE LARKSPUR' in tall condensed capitals across the top, and 'NOW SHOWING TONIGHT' on the line below, framed by rows of round marquee bulbs. The lettering is evenly spaced, high-contrast and sharp. No other text, no brand names, no logos.

Settings

size: 1536x1024
quality: high
complexity: high
background: opaque
moderation: auto
output_format: png
frame_count: 9
frame_duration_ms: 200

A lit cinema marquee on a brick theater facade at night. The marquee's white illuminated board shows black changeable-letter text: 'THE LARKSPUR' in tall conde… — Output from GPT Image 2.6 Sunburst

API request

# pip install "segmind>=1.1.0"
# export SEGMIND_API_KEY="YOUR_API_KEY"
import segmind

result = segmind.run(
    "gpt-image-2.6-sunburst",
    prompt="A lit cinema marquee on a brick theater facade at night. The marquee's white illuminated board shows black changeable-letter text: 'THE LARKSPUR' in tall condensed capitals across the top, and 'NOW SHOWING TONIGHT' on the line below, framed by rows of round marquee bulbs. The lettering is evenly spaced, high-contrast and sharp. No other text, no brand names, no logos.",
    size="1536x1024",
    quality="high",
    complexity="high",
    background="opaque",
    moderation="auto",
    output_format="png",
    frame_count=9,
    frame_duration_ms=200,
    output_compression=100,
)
print(result.get("output"))

# 1. Submit the request; the reply carries its request_id
REQUEST_ID=$(curl -s -X POST "https://api.segmind.com/v2/gpt-image-2.6-sunburst" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "A lit cinema marquee on a brick theater facade at night. The marquee'\''s white illuminated board shows black changeable-letter text: '\''THE LARKSPUR'\'' in tall condensed capitals across the top, and '\''NOW SHOWING TONIGHT'\'' on the line below, framed by rows of round marquee bulbs. The lettering is evenly spaced, high-contrast and sharp. No other text, no brand names, no logos.",
  "size": "1536x1024",
  "quality": "high",
  "complexity": "high",
  "background": "opaque",
  "moderation": "auto",
  "output_format": "png",
  "frame_count": 9,
  "frame_duration_ms": 200,
  "output_compression": 100
}' | jq -r .request_id)

# 2. Check its status until it is COMPLETED
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID/status" \
  -H "x-api-key: YOUR_API_KEY"

# 3. Fetch the result
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID" \
  -H "x-api-key: YOUR_API_KEY"

How to make a transparent-background dessert cutout with GPT Image 2.6 Sunburst

Prompt

A neat stack of five French macarons in pastel colors - pistachio green, rose pink, lavender, lemon yellow and pale blue - each with a smooth domed shell, a ruffled 'foot' and a thick cream filling, centered with generous padding. Clean sharp alpha edges, fully transparent background, no surface, no backdrop, no shadow, no checkerboard, no text, no logos.

Settings

size: 1024x1024
quality: high
complexity: high
background: transparent
moderation: auto
output_format: png
frame_count: 9
frame_duration_ms: 200

A neat stack of five French macarons in pastel colors - pistachio green, rose pink, lavender, lemon yellow and pale blue - each with a smooth domed shell, a ru… — Output from GPT Image 2.6 Sunburst

API request

# pip install "segmind>=1.1.0"
# export SEGMIND_API_KEY="YOUR_API_KEY"
import segmind

result = segmind.run(
    "gpt-image-2.6-sunburst",
    prompt="A neat stack of five French macarons in pastel colors - pistachio green, rose pink, lavender, lemon yellow and pale blue - each with a smooth domed shell, a ruffled 'foot' and a thick cream filling, centered with generous padding. Clean sharp alpha edges, fully transparent background, no surface, no backdrop, no shadow, no checkerboard, no text, no logos.",
    size="1024x1024",
    quality="high",
    complexity="high",
    background="transparent",
    moderation="auto",
    output_format="png",
    frame_count=9,
    frame_duration_ms=200,
    output_compression=100,
)
print(result.get("output"))

# 1. Submit the request; the reply carries its request_id
REQUEST_ID=$(curl -s -X POST "https://api.segmind.com/v2/gpt-image-2.6-sunburst" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "A neat stack of five French macarons in pastel colors - pistachio green, rose pink, lavender, lemon yellow and pale blue - each with a smooth domed shell, a ruffled '\''foot'\'' and a thick cream filling, centered with generous padding. Clean sharp alpha edges, fully transparent background, no surface, no backdrop, no shadow, no checkerboard, no text, no logos.",
  "size": "1024x1024",
  "quality": "high",
  "complexity": "high",
  "background": "transparent",
  "moderation": "auto",
  "output_format": "png",
  "frame_count": 9,
  "frame_duration_ms": 200,
  "output_compression": 100
}' | jq -r .request_id)

# 2. Check its status until it is COMPLETED
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID/status" \
  -H "x-api-key: YOUR_API_KEY"

# 3. Fetch the result
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID" \
  -H "x-api-key: YOUR_API_KEY"

How to generate a 360-degree product turntable animation with GPT Image 2.6 Sunburst

Prompt

A vintage twin-lens reflex camera stands on a seamless light-grey backdrop and rotates a smooth, steady 360-degree turntable around its vertical axis, one full revolution across the loop. It is a boxy black-and-chrome body with two stacked round glass lenses, a knurled focus knob, a brushed-metal winding crank and textured black leatherette panels. Keep the lens rings, engraved dial markings, rivets and leather grain identical in every frame; even studio lighting, fixed camera, no zoom. The last frame flows back into the first. No brand names, no logos, no text.

Settings

size: 768x768
quality: high
complexity: high
background: opaque
moderation: auto
output_format: gif
frame_count: 12
frame_duration_ms: 160

A vintage twin-lens reflex camera stands on a seamless light-grey backdrop and rotates a smooth, steady 360-degree turntable around its vertical axis, one full… — Output from GPT Image 2.6 Sunburst

API request

# pip install "segmind>=1.1.0"
# export SEGMIND_API_KEY="YOUR_API_KEY"
import segmind

result = segmind.run(
    "gpt-image-2.6-sunburst",
    prompt="A vintage twin-lens reflex camera stands on a seamless light-grey backdrop and rotates a smooth, steady 360-degree turntable around its vertical axis, one full revolution across the loop. It is a boxy black-and-chrome body with two stacked round glass lenses, a knurled focus knob, a brushed-metal winding crank and textured black leatherette panels. Keep the lens rings, engraved dial markings, rivets and leather grain identical in every frame; even studio lighting, fixed camera, no zoom. The last frame flows back into the first. No brand names, no logos, no text.",
    size="768x768",
    quality="high",
    complexity="high",
    background="opaque",
    moderation="auto",
    output_format="gif",
    frame_count=12,
    frame_duration_ms=160,
    output_compression=100,
)
print(result.get("output"))

# 1. Submit the request; the reply carries its request_id
REQUEST_ID=$(curl -s -X POST "https://api.segmind.com/v2/gpt-image-2.6-sunburst" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "A vintage twin-lens reflex camera stands on a seamless light-grey backdrop and rotates a smooth, steady 360-degree turntable around its vertical axis, one full revolution across the loop. It is a boxy black-and-chrome body with two stacked round glass lenses, a knurled focus knob, a brushed-metal winding crank and textured black leatherette panels. Keep the lens rings, engraved dial markings, rivets and leather grain identical in every frame; even studio lighting, fixed camera, no zoom. The last frame flows back into the first. No brand names, no logos, no text.",
  "size": "768x768",
  "quality": "high",
  "complexity": "high",
  "background": "opaque",
  "moderation": "auto",
  "output_format": "gif",
  "frame_count": 12,
  "frame_duration_ms": 160,
  "output_compression": 100
}' | jq -r .request_id)

# 2. Check its status until it is COMPLETED
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID/status" \
  -H "x-api-key: YOUR_API_KEY"

# 3. Fetch the result
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID" \
  -H "x-api-key: YOUR_API_KEY"