APIMART-12W22W/gpt-image2.5-api-gateway-apimart

image2.5 API Gateway on APIMart — GPT Image 2.5 (gpt-image-2.5 / gptimage2.5) Flare & Sunburst

$0.0085 per 1K image, billed per delivered image — one OpenAI-compatible endpoint at https://api.apimart.ai/v1, no monthly plan required. (observed 2026-09-17)

Get an API key · Live pricing · Model page · ⚡ 60-second quickstart

Why teams call GPT Image 2.5 (gpt-image-2.5-ext) through APIMart

• One key, entire catalog. The same https://api.apimart.ai/v1 base URL and Authorization header reach GPT Image 2.5 (gpt-image-2.5-ext) and 300+ other image, video and language models — switch the model field, not your client.
• $1 minimum, pay as you go. No subscription and no prepaid plan to size up front: top up from $1 and spend it on calls. There is no free quota to burn through first, so the price in this table is the price you pay.
• The charge comes back in the response. Every call reports the amount billed (cost / credits_cost), so a spend number is read per call instead of guessed at month end.
• Async by design. Submit, take the task_id, poll GET /v1/tasks/{id} — batching and retries are ordinary queue work, not a bespoke integration.

A working map of the image2.5 / gpt image2.5 (GPT Image 2.5) API gateway on APIMart: which model ID to call, how the flash-cheap ext route differs from the token-billed official route, and what the output actually looks like. Includes a 12-image prompt gallery, copy-paste cURL / Python / JavaScript calls, and the async task-polling loop.

Name variants you can search for: image2.5 · image 2.5 · gpt image2.5 · gpt image 2.5 · gpt-image-2.5 · gptimage2.5 · gpt-image2.5 · GPT Image 2.5 — all refer to the same OpenAI image family served here as gpt-image-2.5-flare, gpt-image-2.5-sunburst and gpt-image-2.5-ext.

Contents

• Model routes and IDs
• Observed pricing
• What the output looks like
• Quickstart
• Request and response reference
• What you can call today
• FAQ
• Attributed links
• Repository map

Why an image2.5 gateway instead of a direct vendor call

An AI API gateway keeps one base URL, one key and one request shape in front of several image models. For teams shipping image features, the practical wins are boring but real:

1. One credential. A single Authorization: Bearer header reaches image2.5 plus the rest of the catalog, so keys rotate in one place. 2. Flat, predictable unit cost. The relayed gpt-image-2.5-ext route is billed per delivered image, so a content pipeline can estimate cost before it runs. Token-billed routes cannot do that without measuring output size first. 3. Async by default. POST /v1/images/generations returns a task_id; the poll URL is part of the response, which makes queueing and retry logic straightforward. 4. Model fallback. If one variant is saturated, switch version (or the whole model ID) without touching the client.

This repository documents route selection, price expectations and the request/response shapes that were actually observed.

Model routes and IDs

| Route | model value | Selector | Billing style | Best for | | --- | --- | --- | --- | --- | | Official (token) | gpt-image-2.5-flare | n/a | token usage, quality low → max | everyday generation, batch drafts | | Official (token) | gpt-image-2.5-sunburst | n/a | token usage, quality low → max | editing precision, production assets | | Relayed (per image) | gpt-image-2.5-ext | version: "flare" | per delivered image (n ≤ 4) | high-volume generation at a flat price | | Relayed (per image) | gpt-image-2.5-ext | version: "sunburst" | per delivered image (n ≤ 4) | edits and reference-driven work at a flat price |

Both relayed variants accept resolution 1K / 2K / 4K, ten aspect ratios plus auto, and up to 16 reference images in image_urls. The official route adds exact pixel dimensions and the low / medium / high / xhigh / max quality ladder.

• Family: GPT Image 2.5 — the OpenAI image generation and editing series served through APIMart
• Model IDs: gpt-image-2.5-flare, gpt-image-2.5-sunburst (official route); gpt-image-2.5-ext with version: flare|sunburst (per-image relay route)
• Base URL: https://api.apimart.ai/v1 — OpenAI-compatible POST /v1/images/generations
• Async tasks: submit, then poll GET /v1/tasks/{task_id} until status: completed
• Output tiers: 1K, 2K, 4K; up to 16 reference images for image-to-image; n ≤ 4
• Observed 1K price on the relayed route: $0.0085 per delivered image (checked 2026-09-16)

Observed pricing

| version | 1K | 2K | 4K | billing unit | | --- | --- | --- | --- | --- | | flare | $0.0085 | $0.014 | $0.021 | per delivered image | | sunburst | $0.0085 | $0.014 | check live pricing | per delivered image |

What that costs at scale

| Spend | Cost | | --- | --- | | 100 images | $0.85 | | 1000 images | $8.50 | | 10000 images | $85.00 |

Linear at the observed per-unit rate, no volume discount assumed. Snapshot 2026-09-17; re-check the live table before committing a budget.

Per-image billing on the relayed route is charged for delivered images, and the task response reports the exact amount in cost / credits_cost, so the table above can be re-verified after a single paid call. The official gpt-image-2.5-flare / gpt-image-2.5-sunburst route is token-billed with a low → medium → high → xhigh → max quality ladder, which is why this repository keeps both the flat per-image expectation and the token-billed option side by side. Snapshot date: 2026-09-16.

What the output looks like

Every render below came from a single POST /v1/images/generations call on the relayed route, at the aspect ratio shown. | Output | Recipe | Use case | Version | Ratio | Prompt | | --- | --- | --- | --- | --- | --- | | | Cinematic night street | Cinematic still | flare | 16:9 | Rain-slicked Tokyo alley at night, neon sign reflections on wet asphalt, a lone cyclist with an umbrella, cinematic 35mm film still, shallow depth of field | | | Fashion editorial portrait | Fashion | flare | 3:4 | Fashion editorial portrait of a model wearing an oversized ivory trench coat, seamless light grey studio backdrop, crisp high key lighting, medium format detail | | | App icon / 3D asset | UI asset | flare | 1:1 | Glossy 3D app icon of a paper plane folded from a coral to violet gradient material, floating on a soft neutral grey backdrop, subtle contact shadow, centered composition | | | Architectural render | Architecture | flare | 16:9 | Modern cliffside villa at golden hour, infinity pool reflecting the sky, warm interior lights, architectural photography, 24mm perspective, ultra sharp detail | | | Flat vector dashboard panel | Design / infographic | sunburst | 16:9 | Flat vector dashboard panel with three gauge dials, abstract latency curves and clean geometric cards, muted blue and sand palette, generous white space, crisp vector edges | | | Isometric 3D cutaway | 3D illustration | sunburst | 1:1 | Isometric cutaway of a compact smart home control room, tiny mid century furniture, pastel palette, clay render finish, clean soft shadows, 3D illustration | | | Layered paper-cut illustration | Editorial illustration | sunburst | 4:3 | Layered paper cut illustration of a mountain lake sunrise, five depth layers, soft pastel palette, subtle drop shadows, art print composition | | | Cosmetics packshot | Product / cosmetics | sunburst | 1:1 | Cosmetics packshot of a frosted glass bottle with amber serum, water droplets on the surface, seamless pale pink backdrop, high detail commercial product photography | | | Studio product hero | Product / e-commerce | flare | 1:1 | Studio product hero of a matte ceramic espresso cup on a warm stone pedestal, soft window light, 85mm lens look, minimal beige backdrop, crisp shadow | | | Game key art | Game concept art | sunburst | 16:9 | Fantasy game key art, an armored knight standing on a floating stone ruin above a sea of clouds, dramatic backlight, painterly detail, wide cinematic composition | | | Menu food photography | Food photography | flare | 4:3 | Overhead flat lay of a spicy ramen bowl with a soft boiled egg, chopsticks resting on the rim, dark slate table, natural side light, editorial food photography | | | Macro nature detail | Nature macro | sunburst | 3:2 | Macro photograph of a dew covered monarch butterfly wing, iridescent orange scales, deep black background, focus stacked detail, studio lighting |

Every recipe ships with the exact JSON body in examples/.

Quickstart

The relayed route is asynchronous: submit, then poll the task ID.

# text to image on the per-image route
IDEMPOTENCY_KEY="$(uuidgen)"
curl --request POST \
  --url https://api.apimart.ai/v1/images/generations \
  --header "Authorization: Bearer $APIMART_API_KEY" \
  --header 'Content-Type: application/json' \
  --header 'X-APIMart-Response-Version: 2026-07-27' \
  --header "Idempotency-Key: $IDEMPOTENCY_KEY" \
  --data '{
    "model": "gpt-image-2.5-ext",
    "version": "flare",
    "prompt": "A cozy reading nook beside a window on a rainy day, warm table lamp, cinematic lighting",
    "size": "1:1",
    "resolution": "1K",
    "n": 1
  }'

import os, time, uuid, requests
BASE = "https://api.apimart.ai/v1"
HEADERS = {
    "Authorization": f"Bearer {os.environ['APIMART_API_KEY']}",
    "Content-Type": "application/json",
    "X-APIMart-Response-Version": "2026-07-27",
    "Idempotency-Key": str(uuid.uuid4()),   # reuse on retry, not on a new image
}
def generate(prompt: str, version: str = "flare", resolution: str = "1K", size: str = "1:1") -> str:
    r = requests.post(f"{BASE}/images/generations", headers=HEADERS, timeout=60, json={
        "model": "gpt-image-2.5-ext", "version": version, "prompt": prompt,
        "size": size, "resolution": resolution, "n": 1,
    })
    r.raise_for_status()
    task_id = r.json()["data"]["id"]
    while True:
        t = requests.get(f"{BASE}/tasks/{task_id}", headers=HEADERS, timeout=60).json()["data"]
        if t["status"] in ("completed", "failed"):
            return t
        time.sleep(5)

const headers = {
  Authorization: Bearer ${process.env.APIMART_API_KEY},
  "Content-Type": "application/json",
  "X-APIMart-Response-Version": "2026-07-27",
  "Idempotency-Key": crypto.randomUUID(),
};
const res = await fetch("https://api.apimart.ai/v1/images/generations", {
  method: "POST",
  headers,
  body: JSON.stringify({
    model: "gpt-image-2.5-ext", version: "flare", prompt: "A sky garden at dawn, architectural photography",
    size: "16:9", resolution: "1K", n: 1,
  }),
});
const { data } = await res.json();          // data.id === task id
// poll GET https://api.apimart.ai/v1/tasks/${data.id} until data.status === "completed"

Official (token-billed) route, for comparison — same path, no version, quality ladder instead:

curl --request POST --url https://api.apimart.ai/v1/images/generations \
  --header "Authorization: Bearer $APIMART_API_KEY" --header 'Content-Type: application/json' \
  --data '{"model":"gpt-image-2.5-sunburst","prompt":"Preserve the product label, replace the background with soft off-white, add a natural cast shadow","size":"1:1","resolution":"1k","quality":"high","n":1}'

Full field reference: official route docs and ext route docs. Get a key at apimart.ai/keys.

Request and response reference (ext route)

| Field | Type | Default | Notes | | --- | --- | --- | --- | | model | string | required | gpt-image-2.5-ext | | version | string | flare | flare or sunburst | | prompt | string | required | must not be empty after trimming | | resolution | string | 1K | 1K, 2K, 4K | | size | string | auto | auto or 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9 | | n | integer | 1 | 1–4 per request on the relayed route | | image_urls | string[] | — | up to 16 references, URL or data URL, no extra charge |

Submission returns 202 with data.id (the task ID) and data.poll_url; GET /v1/tasks/{task_id} then reports status (pending → processing → completed / failed), progress, cost, credits_cost and, when finished, result.images[].url with an expires_at timestamp. Download outputs before that timestamp — the URLs are temporary.

What you can call today

| Capability | How it is reached | | --- | --- | | Text to image | model: gpt-image-2.5-ext, version: flare or sunburst | | Image to image / edit | same call plus image_urls: [...] (up to 16 references) | | Batch of up to 4 | n: 4 on the relayed route | | Resolution tiers | resolution: 1K \| 2K \| 4K | | Aspect ratio | ten ratios plus auto | | Progress reporting | poll GET /v1/tasks/{task_id} for progress and status |

First-call troubleshooting

| Symptom | Likely cause | Fix | | --- | --- | --- | | 401 / invalid api key | key missing, truncated, or a stray newline pasted into the header | Re-copy it from the console; the header is Authorization: Bearer $APIMART_API_KEY | | balance / credit error | the account has no balance | Top up from $1 in the console — there is no free quota to fall back on | | 429 | concurrent requests on one key | Back off, then retry the same request with the same Idempotency-Key | | 400 / model not found | wrong route for the id: the per-unit alias needs its version, the official id must not send one | Copy the exact model value from the route table above | | task ends failed | prompt rejected by the filter, or a reference image URL expired | Re-submit with a new Idempotency-Key and re-host the reference image | | result URL stops working | result links expire | Download the file as soon as the task reports completed |

FAQ

What is the image2.5 API model ID?

The official route exposes gpt-image-2.5-flare and gpt-image-2.5-sunburst. The relayed per-image route uses gpt-image-2.5-ext and selects the variant with version: "flare" or version: "sunburst".

How much does one image2.5 image cost on the cheap route?

Observed 1K output on gpt-image-2.5-ext billed $0.0085 per delivered image (checked 2026-09-16). 2K and 4K steps are listed on the pricing page; re-check before a large run.

Is the image2.5 API OpenAI-compatible?

The endpoint path and JSON body follow the OpenAI images shape, but this is an APIMart-hosted gateway: the base URL is https://api.apimart.ai/v1, n is capped at 4 on the relayed route, and long jobs return a task ID instead of an inline image.

Do reference images cost extra?

On the relayed route, reference images are documented as incurring no additional charge; billing tracks version, resolution and the number of images actually delivered.

How do I poll a task?

Use the poll_url returned on submission, e.g. GET /v1/tasks/task_01EXAMPLE. A completed task carries result.images[].url plus expires_at, so download results before the window closes.

Related searches

• image2.5 api
• image 2.5 api
• image2.5 api gateway
• image2-5 api
• gpt-image-2.5 api
• image2.5 api pricing
• ai api relay
• ai api gateway
• ai api aggregator
• apimart image2.5
• image2.5 api documentation
• openai compatible image api
• image2 5
• image2 5 api
• image 2 5 api
• apimart
• api relay

---

Start with $1. Get an API key → check live pricing → open GPT Image 2.5 (gpt-image-2.5-ext) in the model library. The first call is three steps: submit, poll task_id, read the charged amount off the response.

Attributed links (how this repository is measured)

Every outbound link in this repository points at APIMart through a short link, so visits coming from this page are attributed instead of arriving as anonymous traffic.

| Purpose | Attributed link | Target | | --- | --- | --- | | Open GPT Image 2.5 on APIMart | | apimart.ai/model/gpt-image-2-5 | | Current APIMart pricing | | apimart.ai/pricing | | Get an API key on APIMart | | apimart.ai/keys |

• [ ] Attribution target: the three go.apimart.ai short links above, all minted through the promo link API (302 with utm_source=kol_sponsor&utm_medium=sponsor&sclid=...). The endpoint docs on docs.apimart.ai are referenced as plain links: the link service only accepts the apimart.ai main domain, so no attributed short link exists for them.
• [ ] Re-check the price on the pricing page before a production run: promotional routing can change.

Disclosure

APIMart is the service described in this repository; this page is published to document it, not to claim official status. The ext route is a third-party relay endpoint billed per delivered image, while the gpt-image-2.5-flare / gpt-image-2.5-sunburst models are the token-billed route. Model names, prices and limits belong to their respective owners, and everything here is observation-dated (2026-09-16). Verify with a single paid request before scaling volume.

Repository map

image2.5-api-gateway-apimart/
  PROMPTS.md           every recipe with its output
  README.md            overview, pricing, quickstart and FAQ
  examples/
    curl.sh            submit + poll with curl
    python_generate.py end-to-end Python client
    javascript.mjs     Node 18+ equivalent
  tools/check_links.py attribution + prompt-data validator
  .github/workflows/validate.yml  CI for the validator
  assets/              example renders (JPEG, resized for the README)
  LICENSE              MIT

License

MIT — see LICENSE. Model names and vendor documentation remain the property of their owners.