GPT Image 2.6 Flare API

GPT Image 2.6 Flare: Text-to-Image and Animated GIF

What is GPT Image 2.6 Flare?

GPT Image 2.6 Flare is OpenAI's speed-first model in the GPT Image 2.6 family, served on Segmind as one endpoint for text-to-image generation, reference-guided editing, and native animated GIF output. Describe a scene and Flare returns a PNG, JPEG or WEBP still; switch the output format to GIF and the same request renders a looping animation, with no separate video model and no frame stitching. Flare is the route tuned for speed and big, sweeping motion, so it is the natural default for iterating on animated social and ad creative.

Key Features

Native animated GIF: 2 to 16 frames (9 by default), adjustable frame duration, and a loop that plays forever.
Big-motion cinemagraphs: crashing surf, billowing smoke, swaying trees, flames and pours.
Fast turnaround: about twice as quick as GPT Image 2.6 Sunburst on matched requests.
Legible in-image text for posters, labels and UI copy, plus edit mode to restyle or animate a reference photo.
Transparent backgrounds with PNG, WEBP or GIF, and quality and complexity controls that trade planning effort for speed.

Best Use Cases

Social and ad creative: feed loops, display banners and email headers, where GIF plays almost everywhere.
Cinemagraphs: one dramatic moving element such as flames, rain, surf or a molten pour.
Rapid motion concepting: try many ideas at low quality, then re-render the winner at high.
Stills and stickers: product shots, posters and mockups with crisp typography, or reactions on transparent backgrounds.

Prompt Tips and Output Quality

Write a GIF prompt in three parts: the subject, the motion, and how the loop closes ("the last frame flows back into the first"). Big, continuous actions read best at thumbnail size. For a GIF, size sets one frame; 960x960 and 1280x720 both hold 9 frames reliably, while very small 16:9 frames can be rejected, so lower the frame count or pick a larger frame if that happens. Loop length is frame count times frame duration, so 9 frames at 200 ms is a 1.8-second loop. Draft at low quality and finish at high.

How to Use GPT Image 2.6 Flare

GPT Image 2.6 Flare generates stills and native animated GIFs from a text prompt, and edits or animates your own images when you add references. It is the fast route, so reach for it to iterate quickly and for animations with big, energetic motion.

For a GIF, set the output format to gif and write the prompt in three parts: the subject, the motion, and the loop ("the last frame flows back into the first"). Size is the size of one frame. Frame count sets the number of frames (2 to 16) and frame duration sets how long each frame stays on screen; loop length is frames times duration.

Recommended Settings

Cinemagraph or social loop: output format gif, size 960x960 or 1280x720, frame count 9, frame duration 200 ms, quality high. The default foundry molten-pour example was rendered end to end at these settings and produced a clean, seamless loop with no visible jump.
Smooth, small loop or sticker: size auto, frame count up to 16, frame duration 80 to 120 ms, background transparent.
Fast motion drafts: quality low, complexity low, then re-render the best idea at quality high.
Animate your own image: add one clean reference and describe only the motion.
Stills with text: output format png, size 1024x1536 or 1536x1024, quality high.
Transparent assets: background transparent with png, webp or gif; jpeg always stays opaque.

If a GIF request is rejected, the frames do not fit the model's limits (very small 16:9 frames such as 640x360 are a known case). Lower the frame count or pick a larger frame size. For 360-degree rotations and turntables where fine detail must match in every frame, use GPT Image 2.6 Sunburst.

Examples to try

Each example below ran end to end through GPT Image 2.6 Flare at the settings shown on its card — the same prompt, size and quality, plus frame count and duration for the GIFs — so replaying a card reproduces the loop or still you see.

Native animated GIF, big motion (default). Prompt: "A dark iron foundry: a heavy steel ladle tips and a glowing stream of molten metal pours into a sand mould below, a continuous fountain of bright orange sparks spitting off the falling stream and skittering across the cast-iron floor, heat haze rippling above the metal, deep shadows on the sooty brick walls behind. The pour runs steadily and the shower of sparks keeps cascading, so the last frame flows back into the first for a seamless loop. Gritty industrial realism, hard rim light from the molten glow. No people, no text, no logos, no brand marks, no readable signage." A ladle of molten metal pours and a shower of orange sparks cascades across the frame in a seamless nine-frame loop. Write the prompt in three parts — subject, motion, and how the loop closes — and keep the output format on gif.
Legible in-image text (still PNG). Prompt: "A hand-stitched cross-stitch sampler stretched in a round pale-birch embroidery hoop on natural flax linen. A top band spells the full alphabet in even red block capitals; below it two centered lines read "HOME SWEET HOME" and "EST. 1887" in navy stitched serif letters, framed by a border of small cross-stitched daisies and strawberries. Flat overhead view, crisp even studio light. No brand names, no logos, no trademarks." Every letter of the alphabet band and the "HOME SWEET HOME / EST. 1887" motto reads cleanly in the hoop. Switch the output format to png and quality to high, and spell the exact words you want in quotes.
Transparent animated sticker. Prompt: "A playful looping sticker of a chunky retro wind-up tin robot with round rivets, a square head and a little antenna, drawn as a glossy flat illustration with a clean thick white die-cut border. The robot waddles from side to side on the spot and the round lamp on its chest blinks on and off, the motion repeating so the last frame flows back into the first for a seamless loop. Fully transparent background with a real alpha channel, no backdrop, no text, no logos, no brand marks." The tin robot bobs with a blinking chest lamp on a see-through background, ready to drop onto any layout. Set the background to transparent with png, webp or gif — jpeg always stays opaque.

Reference-guided animation

Flare switches to edit mode the moment you add an image to the reference images (image_urls): it takes one clean photo, and you describe only the motion or restyle you want.

Animate your own photo into a cinemagraph. Prompt: "Animate this exact bowl of ramen as a cinemagraph: keep the bowl, noodles, egg, chopsticks and table perfectly still, and add soft wisps of steam curling and drifting upward from the hot broth. The steam rises and thins in a continuous cycle so the last frame flows back into the first for a seamless loop. Photoreal, no text, no logos, no brand marks." (run over a supplied photo of a bowl of ramen). The bowl holds still while steam rises and curls off the broth in a loop. Add one clean reference and keep the rest of the frame still so only the motion you name moves.

More prompts this route is built for: "a paper lantern swaying on a transparent background, seamless loop", "rain streaking down a dark window, everything else held still", and "a flag rippling in a steady wind, the last frame flowing back into the first".

GPT Image 2.6 Flare examples

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

How to make a looping animation of a molten metal foundry pour with GPT Image 2.6 Flare

Prompt

A dark iron foundry: a heavy steel ladle tips and a glowing stream of molten metal pours into a sand mould below, a continuous fountain of bright orange sparks spitting off the falling stream and skittering across the cast-iron floor, heat haze rippling above the metal, deep shadows on the sooty brick walls behind. The pour runs steadily and the shower of sparks keeps cascading, so the last frame flows back into the first for a seamless loop. Gritty industrial realism, hard rim light from the molten glow. No people, no text, no logos, no brand marks, no readable signage.

Settings

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

A dark iron foundry: a heavy steel ladle tips and a glowing stream of molten metal pours into a sand mould below, a continuous fountain of bright orange sparks… — Output from GPT Image 2.6 Flare

API request

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

result = segmind.run(
    "gpt-image-2.6-flare",
    prompt="A dark iron foundry: a heavy steel ladle tips and a glowing stream of molten metal pours into a sand mould below, a continuous fountain of bright orange sparks spitting off the falling stream and skittering across the cast-iron floor, heat haze rippling above the metal, deep shadows on the sooty brick walls behind. The pour runs steadily and the shower of sparks keeps cascading, so the last frame flows back into the first for a seamless loop. Gritty industrial realism, hard rim light from the molten glow. No people, no text, no logos, no brand marks, no readable signage.",
    image_urls=[],
    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-flare" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "A dark iron foundry: a heavy steel ladle tips and a glowing stream of molten metal pours into a sand mould below, a continuous fountain of bright orange sparks spitting off the falling stream and skittering across the cast-iron floor, heat haze rippling above the metal, deep shadows on the sooty brick walls behind. The pour runs steadily and the shower of sparks keeps cascading, so the last frame flows back into the first for a seamless loop. Gritty industrial realism, hard rim light from the molten glow. No people, no text, no logos, no brand marks, no readable signage.",
  "image_urls": [],
  "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 design a cross-stitch sampler mockup for a craft shop with GPT Image 2.6 Flare

Prompt

A hand-stitched cross-stitch sampler stretched in a round pale-birch embroidery hoop on natural flax linen. A top band spells the full alphabet in even red block capitals; below it two centered lines read "HOME SWEET HOME" and "EST. 1887" in navy stitched serif letters, framed by a border of small cross-stitched daisies and strawberries. Flat overhead view, crisp even studio light. No brand names, no logos, no trademarks.

Settings

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

A hand-stitched cross-stitch sampler stretched in a round pale-birch embroidery hoop on natural flax linen. A top band spells the full alphabet in even red blo… — Output from GPT Image 2.6 Flare

API request

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

result = segmind.run(
    "gpt-image-2.6-flare",
    prompt="A hand-stitched cross-stitch sampler stretched in a round pale-birch embroidery hoop on natural flax linen. A top band spells the full alphabet in even red block capitals; below it two centered lines read \"HOME SWEET HOME\" and \"EST. 1887\" in navy stitched serif letters, framed by a border of small cross-stitched daisies and strawberries. Flat overhead view, crisp even studio light. No brand names, no logos, no trademarks.",
    image_urls=[],
    size="960x960",
    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-flare" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "A hand-stitched cross-stitch sampler stretched in a round pale-birch embroidery hoop on natural flax linen. A top band spells the full alphabet in even red block capitals; below it two centered lines read \"HOME SWEET HOME\" and \"EST. 1887\" in navy stitched serif letters, framed by a border of small cross-stitched daisies and strawberries. Flat overhead view, crisp even studio light. No brand names, no logos, no trademarks.",
  "image_urls": [],
  "size": "960x960",
  "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 turn a ramen photo into a steaming cinemagraph loop with GPT Image 2.6 Flare

Prompt

Animate this exact bowl of ramen as a cinemagraph: keep the bowl, noodles, egg, chopsticks and table perfectly still, and add soft wisps of steam curling and drifting upward from the hot broth. The steam rises and thins in a continuous cycle so the last frame flows back into the first for a seamless loop. Photoreal, no text, no logos, no brand marks.

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 bowl of ramen as a cinemagraph: keep the bowl, noodles, egg, chopsticks and table perfectly still, and add soft wisps of steam curling and d… — Output from GPT Image 2.6 Flare

API request

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

result = segmind.run(
    "gpt-image-2.6-flare",
    prompt="Animate this exact bowl of ramen as a cinemagraph: keep the bowl, noodles, egg, chopsticks and table perfectly still, and add soft wisps of steam curling and drifting upward from the hot broth. The steam rises and thins in a continuous cycle so the last frame flows back into the first for a seamless loop. Photoreal, no text, no logos, no brand marks.",
    image_urls=[
        "https://segmind-resources.s3.amazonaws.com/input/gpt-image-2.6-flare-example-reference-animate-c6fa1a78.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-flare" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "Animate this exact bowl of ramen as a cinemagraph: keep the bowl, noodles, egg, chopsticks and table perfectly still, and add soft wisps of steam curling and drifting upward from the hot broth. The steam rises and thins in a continuous cycle so the last frame flows back into the first for a seamless loop. Photoreal, no text, no logos, no brand marks.",
  "image_urls": [
    "https://segmind-resources.s3.amazonaws.com/input/gpt-image-2.6-flare-example-reference-animate-c6fa1a78.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 make an animated transparent sticker of a tin robot with GPT Image 2.6 Flare

Prompt

A playful looping sticker of a chunky retro wind-up tin robot with round rivets, a square head and a little antenna, drawn as a glossy flat illustration with a clean thick white die-cut border. The robot waddles from side to side on the spot and the round lamp on its chest blinks on and off, the motion repeating so the last frame flows back into the first for a seamless loop. Fully transparent background with a real alpha channel, no backdrop, no text, no logos, no brand marks.

Settings

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

A playful looping sticker of a chunky retro wind-up tin robot with round rivets, a square head and a little antenna, drawn as a glossy flat illustration with a… — Output from GPT Image 2.6 Flare

API request

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

result = segmind.run(
    "gpt-image-2.6-flare",
    prompt="A playful looping sticker of a chunky retro wind-up tin robot with round rivets, a square head and a little antenna, drawn as a glossy flat illustration with a clean thick white die-cut border. The robot waddles from side to side on the spot and the round lamp on its chest blinks on and off, the motion repeating so the last frame flows back into the first for a seamless loop. Fully transparent background with a real alpha channel, no backdrop, no text, no logos, no brand marks.",
    image_urls=[],
    size="960x960",
    quality="high",
    complexity="high",
    background="transparent",
    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-flare" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "prompt": "A playful looping sticker of a chunky retro wind-up tin robot with round rivets, a square head and a little antenna, drawn as a glossy flat illustration with a clean thick white die-cut border. The robot waddles from side to side on the spot and the round lamp on its chest blinks on and off, the motion repeating so the last frame flows back into the first for a seamless loop. Fully transparent background with a real alpha channel, no backdrop, no text, no logos, no brand marks.",
  "image_urls": [],
  "size": "960x960",
  "quality": "high",
  "complexity": "high",
  "background": "transparent",
  "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"

Can GPT Image 2.6 Flare make animated GIFs?

Yes. Set the output format to gif and it returns a looping GIF directly.

Should I use Flare or Sunburst?

Choose Flare for speed and dramatic motion; choose GPT Image 2.6 Sunburst for 360-degree rotations and fine detail that must stay consistent across frames.

Can I animate my own image?

Yes. Add it to the reference images and describe the motion you want.

Does it support transparent backgrounds?

Yes, with PNG, WEBP or GIF output.

How many frames can a GIF have?

From 2 to 16, with 9 as the default.

GPT Image 2.6 Flare

GPT Image 2.6 Flare: Text-to-Image and Animated GIF

What is GPT Image 2.6 Flare?

Key Features

Best Use Cases

Prompt Tips and Output Quality

How to Use GPT Image 2.6 Flare

Recommended Settings

Examples to try

Reference-guided animation

How to make a looping animation of a molten metal foundry pour with GPT Image 2.6 Flare

How to design a cross-stitch sampler mockup for a craft shop with GPT Image 2.6 Flare

How to turn a ramen photo into a steaming cinemagraph loop with GPT Image 2.6 Flare

How to make an animated transparent sticker of a tin robot with GPT Image 2.6 Flare

Can GPT Image 2.6 Flare make animated GIFs?

Should I use Flare or Sunburst?

Can I animate my own image?

Does it support transparent backgrounds?

How many frames can a GIF have?

Related Models

GPT Image 2

GPT Image 2.5 Sunburst

GPT Image 2.5 Flare

GPT Image 2.6 Sunburst