Create image
Last updated November 23, 2026
Generate images with openai/gpt-image-2.6-flare over the OpenAI-compatible endpoint with SVG, animated GIF, and complexity controls.
Text-to-image generation with openai/gpt-image-2.6-flare over the OpenAI-compatible endpoint. Official OpenAI SDKs work unchanged once base_url points to the gateway.
Flare is the latency-optimized variant of the GPT-Image 2.6 series, delivering the 2.6 additions (svg vector output, animated gif with frame controls, and complexity) at significantly lower end-to-end latency. When detail fidelity and complex composition take priority over speed, use gpt-image-2.6-sunburst.
https://gateway.amux.ai/v1/images/generationsAuthorization
headerAuthorizationstringRequiredBearer <your Amux AI key> Amux AI API key (Bearer token).
Request
application/jsonmodelstringRequiredModel identifier (openai/gpt-image-2.6-flare), optionally suffixed with :amux-special or :adobe to pin a provider (:openai and :azure are coming soon).
promptstringRequiredText prompt describing the desired image (up to 32,000 characters).
nintegerDefault 1Number of images to generate (1–10, default 1). Must be 1 when output_format is svg or when stream is true.
sizestringExact pixel dimensions or auto (default). Standard tiers include 1024x1024 (1:1), 1536x1024 / 1024x1536 (3:2 / 2:3), 1024x768 / 768x1024 (4:3 / 3:4), 2048x2048 (2K 1:1), 2048x1152 (16:9), 2048x1536 / 1536x2048 (2K 4:3 / 3:4), 3840x2160 / 2160x3840 (4K 16:9 / 9:16), and 3264x2448 / 2448x3264 (4K 4:3 / 3:4).
Custom pixel dimensions must satisfy: both edges multiples of 16, long edge ≤ 3840, aspect ratio ≤ 3:1, and total pixels between 655,360 and 8,294,400.
Provider notes: adobe renders true native 4K and exact custom pixel sizes (strictly enforcing ≥ 655,360 total pixels; sizes like 512x512 are rejected); amux-special scales non-4K sizes proportionally to ~1.57 MP (e.g. 1672x941 for 16:9, 1254x1254 for 1:1) and applies post-processed super-resolution for 4K (3840x2160).
aspect_ratiostringAspect ratio (1:1, 4:3, 3:4, 3:2, 2:3, 16:9, or 9:16) as an alternative to size. Prefer size for exact pixel dimensions; if both are provided, size takes precedence and any conversion or omission is noted in amux.notes.
image_sizestringResolution tier (1K, 2K, or 4K), used in combination with aspect_ratio under the same conversion rules.
quality"low" | "medium" | "high" | "xhigh" | "max" | "auto"Default "auto"Render quality tier (low, medium, high, xhigh, max, or auto). Supported on both amux-special and adobe.
complexity"low" | "medium" | "high"Default "high"New in GPT-Image 2.6: controls visual and compositional complexity (low, medium, or high, default high). Supported on both amux-special and adobe.
background"transparent" | "opaque" | "auto"Default "auto"Background mode. Setting transparent requires output_format to be png, webp, or gif (jpeg has no alpha channel).
output_format"png" | "jpeg" | "webp" | "svg" | "gif"Default "png"Output image encoding format (defaults to png). GPT-Image 2.6 adds svg (vector output; generation only, n=1) and gif (animated GIF; generation only); neither supports streaming.
Provider support: Unsupported on adobe (adobe accepts png, jpeg, and webp only). Without a pinned provider, svg and gif requests automatically route to amux-special; pinning :adobe returns 400.
output_compressionintegerDefault 100Compression level (0–100, default 100); applies only to jpeg and webp.
frame_countintegerDefault 9Number of frames in an animated GIF (2–16, default 9). Effective only on generation requests with output_format: "gif"; ignored and noted in amux.notes otherwise.
Provider support: Unsupported on adobe (because gif is unsupported); supported on amux-special.
frame_duration_msintegerDefault 200Duration per GIF frame in milliseconds (≥ 20 and a multiple of 10, default 200). Effective only on generation requests with output_format: "gif"; ignored and noted in amux.notes otherwise.
Provider support: Unsupported on adobe (because gif is unsupported); supported on amux-special.
moderation"auto" | "low"Default "auto"Content moderation strictness (supported on the generation endpoint only).
userstringUnique end-user identifier for abuse detection and monitoring. Omit when unused.
streambooleanDefault falseEnable Server-Sent Events streaming. Requires n: 1 and cannot be combined with output_format: "svg" or "gif".
Provider support: Unsupported on both live providers (amux-special and adobe)—passing stream: true returns 400 until official openai / azure providers launch.
partial_imagesintegerDefault 0Maximum number of low-resolution preview frames emitted during streaming (0–3, default 0). Cannot be combined with svg or gif.
Provider support: Depends on streaming; unsupported on both amux-special and adobe until official openai / azure providers launch.
Response
200responseImage generation succeeded. data and usage match OpenAI's schema; amux contains gateway extensions.
400responseInvalid request parameters (including svg/gif constraint violations, pinning a provider that does not support svg/gif/stream, or n > 1 while streaming).
402responseInsufficient available workspace balance to cover the estimated reservation for this request.
503responseService temporarily unavailable, or the requested model or size tier is currently unavailable.
Providers & availability
openai/gpt-image-2.6-flare is planned across four providers. Append :provider to model (for example, openai/gpt-image-2.6-flare:adobe) to pin a specific provider, or omit the suffix to let routing select based on strategy and parameter support. See the model page for live rates.
| Provider | Status | Billing & tier | 4K & resolution behavior | Unsupported parameters |
|---|---|---|---|---|
amux-special | Live | Value per-image billing | Non-4K sizes scale proportionally to ~1.57 MP (e.g. 1672x941 for 16:9, 1254x1254 for 1:1); 4K (3840x2160) uses post-processed super-resolution | stream, partial_images |
adobe | Live | Value per-image billing (slightly above amux-special) | True native 4K (3840x2160) and exact custom pixels; strictly enforces ≥ 655,360 total pixels | output_format: svg / gif, frame_count, frame_duration_ms, mask, markup, stream, partial_images |
openai | Coming soon | Official token-based billing | Native official full spec | — (full 2.6 spec ready; activates upon launch) |
azure | Coming soon | Official Azure OpenAI token-based billing | Native official full spec | — (Azure deployment endpoints ready; activates upon launch) |
Automatic capability routing: Without an explicit
:providersuffix, requests specifyingoutput_format: "svg",output_format: "gif",mask, ormarkupautomatically skipadobeand route toamux-special. Pinning:adobewith any of those features fails fast with400.
New in 2.6
complexity
Accepts low, medium, or high (default high) to tune compositional density and visual detail independently of quality. Supported on both amux-special and adobe.
output_format: "svg" and "gif"
svg: Generates editable vector graphics. Supported on generation only (not edits), requiresn = 1, and cannot be combined withstream. Hosted.svglinks inamux.imagesare served withContent-Disposition: attachmentfor browser safety.gif: Generates looping animations. Supported on generation only and cannot be combined withstream. Configure frame count withframe_count(2–16, default9) and per-frame timing withframe_duration_ms(≥ 20 and a multiple of 10, default200).- Provider note:
adobedoes not supportsvg,gif,frame_count, orframe_duration_ms. Omit:adobe(or pin:amux-special) when generating SVG or GIF outputs.
Dimensions
Standard pixel tiers:
| Aspect ratio | 1K | 2K | 4K |
|---|---|---|---|
| 1:1 | 1024x1024 | 2048x2048 | — |
| 3:2 / 2:3 | 1536x1024 / 1024x1536 | — | — |
| 4:3 / 3:4 | 1024x768 / 768x1024 | 2048x1536 / 1536x2048 | 3264x2448 / 2448x3264 |
| 16:9 / 9:16 | — | 2048x1152 | 3840x2160 / 2160x3840 |
Custom pixel dimensions are accepted when both edges are multiples of 16, the long edge is ≤ 3840, the aspect ratio is ≤ 3:1, and total pixels fall between 655,360 and 8,294,400. Pin :adobe when true native 4K pixels are required; use :amux-special for cost-optimized standard renders, svg, or gif.
Artifact retention
data[].b64_json contains the Base64-encoded output for the call, and amux.images provides temporary direct URLs. Hosted artifacts are purged after the retention period and return 404 thereafter; download any files you need to retain.
Streaming
Neither live provider (amux-special or adobe) currently supports stream: true or partial_images—requests passing stream: true fail fast with 400 rather than silently downgrading. Streaming previews will activate once official openai / azure providers launch.
Errors
Error responses follow OpenAI's standard schema. See Errors and retries for type codes and retry guidance.
curl https://gateway.amux.ai/v1/images/generations \
-H "Authorization: Bearer $AMUX_API_KEY" \
-d '{
"model": "openai/gpt-image-2.6-flare",
"prompt": "A red fox sitting in snow, photorealistic",
"size": "1024x1024"
}'{
"created": 1786000000,
"size": "<string>",
"quality": "<string>",
"background": "<string>",
"output_format": "<string>",
"data": [
{
"b64_json": "<string>",
"revised_prompt": "<string>"
}
],
"usage": {
"input_tokens": 0,
"output_tokens": 0,
"total_tokens": 0
},
"amux": {
"task_id": "<string>",
"generation_id": "<string>",
"provider": "<string>",
"cost_nano": "<string>",
"images": [
{
"index": 0,
"url": "<string>"
}
],
"notes": [
"<string>"
]
}
}