Dala Futhi Hlela Izithombe Nge-GPT-Image-2.5 Usebenzisa i-APIMart

Okwenziwa i-image2.5 APIMart gateway

I-image2.5 gateway inikeza ukufinyelela emndenini wokukhiqiza nokuhlela we-GPT Image 2.5 nge-APIMart API. Esikhundleni sokuhlanganisa imininingwane yokungena nefomethi yesicelo ehlukile kuyo yonke imodeli yesithombe, usebenzisa i-base URL eyodwa, ithokheni ye-bearer eyodwa, kanye ne-endpoint ye-OpenAI-style /v1/images/generations.

I-API isebenza ngokungavumelani ngesikhathi esisodwa. Isicelo sokukhiqiza sibuyisa i-task ID esikhundleni sesithombe esifakwe ngokuqondile. Uhlelo lwakho bese luhlola i-task endpoint kuze kuqedwe noma kuhluleke ukucutshungulwa. Ama-task aphumelele aqukethe ama-URL esikhashana emiphumela, okufanele alandwe ngaphambi kokuphelelwa yisikhathi.

APIMart iyisevisi ebhalwe kumadokhumenti e-repository. I-endpoint ethi gpt-image-2.5-ext iyindlela yokudlulisa evela kwelinye iqembu, kanti gpt-image-2.5-flare kanye gpt-image-2.5-sunburst yizindlela ezikhokhiswa ngamathokheni. Qinisekisa imikhawulo nezintengo zamanje ngaphambi kokusebenzisa ohlelweni lokukhiqiza.

Izici ezibalulekile

• I-API base URL eyodwa: Izicelo zisebenzisa https://api.apimart.ai/v1.
• Izinhlobo ezimbili zamamodeli: Khetha i-Flare ekukhiqizeni kwansuku zonke kanye nemidwebo yeqoqo, noma i-Sunburst ekunembeni kokuhlela nasezimpahleni ezilungele ukukhiqizwa.
• Izindlela ezimbili zokukhokha: Indlela edlulisiwe ikhokhiswa ngesithombe ngasinye esilethiwe, kanti izindlela ezisemthethweni zisebenzisa ukukhokhiswa ngamathokheni.
• Ukucutshungulwa okungavumelani ngesikhathi esisodwa: Thumela isicelo bese uhlola GET /v1/tasks/{task_id} ukuze ubone inqubekelaphambili.
• Amaleveli amaningi emiphumela: Indlela edlulisiwe isekela 1K, 2K, kanye 4K.
• Ukuqanjwa okuguquguqukayo: Khetha kuma-aspect ratio ayishumi noma usebenzise auto.
• Ukukhiqiza okuqhutshwa yizithenjwa: Faka okungafika kokungu-16 ku-image_urls emsebenzini wokuguqula isithombe ube esinye noma wokuhlela.
• Amabhetshi amancane: Cela izithombe ezilethiwe eziphakathi kwesisodwa nezine usebenzisa u-n endleleni edlulisiwe.

Khetha indlela yemodeli efanele

Indlela edlulisiwe ekhokhiswa ngesithombe

Sebenzisa model: "gpt-image-2.5-ext" uma ufuna ukukhokhiswa okuyisilinganiso esifanayo ngesithombe ngasinye esilethiwe. Khetha uhlobo oluyisisekelo ngo-version: "flare" noma version: "sunburst".

• flare yenzelwe ukukhiqiza ngobuningi, izithombe zansuku zonke, kanye nemidwebo.
• sunburst yenzelwe ukuhlela nomsebenzi wokukhiqiza oqhutshwa yizithenjwa.

I-repository yaphawula intengo engu-$0.0085 ngesithombe esilethiwe esingu-1K kuzo zombili izinhlobo ngo-September 16, 2026. Yaphinde yaqopha u-$0.014 ku-2K, u-$0.021 ku-4K Flare, futhi yancoma ukuhlola intengo yamanje ye-4K Sunburst. Lezi yizibalo ezabhekwa ngalolo suku, akuzona iziqinisekiso zaphakade.

Izindlela ezisemthethweni ezikhokhiswa ngamathokheni

Sebenzisa gpt-image-2.5-flare noma gpt-image-2.5-sunburst ngokuqondile ukuze ukhokhe ngamathokheni. Lezi zindlela azinayo inkambu ye-version futhi zisekela amazinga ekhwalithi e-low, medium, high, xhigh, kanye max. Indlela esemthethweni iphinde isekele ubukhulu obuqondile ngamaphikseli.

Umthetho osebenzayo uwukusebenzisa indlela edlulisiwe lapho izindleko ezibikezelwayo zesithombe ngasinye zibalulekile, bese ucabangela indlela esemthethweni lapho kudingeka izilawuli zekhwalithi noma ubukhulu obuqondile.

Ububanzi bemiphumela eyisibonelo

I-repository ibonisa iziqondiso zokuthwebula izithombe zemikhiqizo, izithombe zabantu, izakhiwo, izimpahla ze-UI, imifanekiso, ubuciko bemidlalo, izithombe zokudla, kanye nezithombe eziseduze kakhulu. Izithombe ezilandelayo ezinikeziwe zibonisa uhlobo lomphumela obonakalayo ofakwe kuphrojekthi.

Ukufakwa nokusethwa

1. Dala ukhiye we-API

Thola ukhiye ekhasini le-API key le-APIMart. Wugcine ku-environment variable esikhundleni sokuwufaka ngokuqondile kukhodi yomthombo.

export APIMART_API_KEY="your-api-key"

2. Khetha iklayenti

I-cURL ayidingi okunye ukuncika kwephrojekthi ngaphandle kokufakwa okujwayelekile komugqa womyalo. Esibonelweni se-Python, faka iphakheji ye-requests:

python -m pip install requests

I-repository iphinde ibe nesibonelo se-JavaScript se-Node.js 18 noma entsha, lapho i-fetch itholakala khona emhlabeni wonke.

3. Qonda ama-header esicelo

Thumela ukhiye we-API ku-Authorization header. Izibonelo zendlela edlulisiwe zisebenzisa ne-response-version header kanye nokhiye oyingqayizivele we-idempotency.

• Authorization: Bearer $APIMART_API_KEY
• Content-Type: application/json
• X-APIMart-Response-Version: 2026-07-27
• Idempotency-Key: unique-request-value

Sebenzisa kabusha i-idempotency key efanayo lapho uzama kabusha ukuthumela okufanayo. Dala ukhiye omusha lapho ufuna ngamabomu i-task entsha yesithombe.

Ukusetshenziswa okuyisisekelo: khiqiza isithombe nge-cURL

Isicelo esilandelayo sisebenzisa indlela ekhokhiswa ngesithombe, sikhetha i-Flare, futhi sicela isithombe esisodwa esiyisikwele esingu-1K:

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
  }'

Ukuthunyelwa okuyimpumelelo kubuyisa i-HTTP 202. Funda isihlonzi somsebenzi ku-data.id noma usebenzise i-data.poll_url ebuyisiwe.

Hlola umsebenzi

Faka esikhundleni se-task_01EXAMPLE nge-ID yomsebenzi ebuyisiwe:

curl --request GET \
  --url https://api.apimart.ai/v1/tasks/task_01EXAMPLE \
  --header "Authorization: Bearer $APIMART_API_KEY" \
  --header "X-APIMart-Response-Version: 2026-07-27"

Isimo somsebenzi sidlula ku-pending naku-processing ngaphambi kokufinyelela ku-completed noma ku-failed. Impendulo ingaphinda ibike i-progress, i-cost, ne-credits_cost. Uma usuqediwe, thola i-URL yesithombe ku-result.images[].url bese uhlola i-expires_at.

Isibonelo se-Python esiphelele

Lesi sibonelo sithumela isithombe esisodwa bese sihlola umsebenzi njalo ngemva kwemizuzwana emihlanu kuze kufinyelele esimweni sokugcina:

import os
import time
import uuid
import requests

BASE_URL = "https://api.apimart.ai/v1"
COMMON_HEADERS = {
    "Authorization": f"Bearer {os.environ['APIMART_API_KEY']}",
    "Content-Type": "application/json",
    "X-APIMart-Response-Version": "2026-07-27",
}

def generate_image(prompt, version="flare", resolution="1K", size="1:1"):
    submit_headers = {
        **COMMON_HEADERS,
        "Idempotency-Key": str(uuid.uuid4()),
    }

    response = requests.post(
        f"{BASE_URL}/images/generations",
        headers=submit_headers,
        json={
            "model": "gpt-image-2.5-ext",
            "version": version,
            "prompt": prompt,
            "size": size,
            "resolution": resolution,
            "n": 1,
        },
        timeout=60,
    )
    response.raise_for_status()
    task_id = response.json()["data"]["id"]

    while True:
        task_response = requests.get(
            f"{BASE_URL}/tasks/{task_id}",
            headers=COMMON_HEADERS,
            timeout=60,
        )
        task_response.raise_for_status()
        task = task_response.json()["data"]

        if task["status"] == "completed":
            return task
        if task["status"] == "failed":
            raise RuntimeError(f"Generation failed: {task}")

        time.sleep(5)

result = generate_image(
    "A sky garden at dawn, architectural photography",
    version="flare",
    resolution="1K",
    size="16:9",
)

for image in result["result"]["images"]:
    print(image["url"], image.get("expires_at"))

Kuklayenti lokukhiqiza, buyisa noma gcina yonke into yomsebenzi ukuze ukwazi ukugcina ulwazi lwezindleko, inqubekelaphambili, ukuphelelwa yisikhathi, namaphutha esikhundleni sokubuyisa i-URL kuphela.

Sebenzisa izithombe eziyisethenjwa ukuhlela

Ukuze wenze ukukhiqizwa kwe-image-to-image noma ukuhlela emzileni odlulisiwe, engeza uhlu lwe-image_urls. Lungaba nama-URL noma ama-data URL afinyelela kwangu-16. Indawo yokugcina ithi izithombe eziyisethenjwa azikhokhiswa imali ehlukile kulo mzila.

{
  "model": "gpt-image-2.5-ext",
  "version": "sunburst",
  "prompt": "Preserve the product label, replace the background with soft off-white, and add a natural cast shadow",
  "image_urls": [
    "https://example.com/source-product.jpg"
  ],
  "size": "1:1",
  "resolution": "2K",
  "n": 1
}

I-Sunburst iwuhlobo oluphakanyiswa yindawo yokugcina ngokunemba kokuhlela nomsebenzi oqondiswa yizithombe eziyisethenjwa. Sebenzisa ama-URL afinyelelekayo noma ama-data URL avumelekile, futhi ugcine uhlu lungaphakathi komkhawulo obhaliwe.

Inkomba yezinkambu zesicelo somzila odlulisiwe

• model: Iyadingeka. Yibeke ku-gpt-image-2.5-ext.
• version: Yamukela i-flare noma i-sunburst; okuzenzakalelayo yi-Flare.
• prompt: Iyadingeka futhi akumele ingabi nalutho ngemva kokususa izikhala ekuqaleni nasekupheleni.
• resolution: Yamukela i-1K, 2K, noma 4K; okuzenzakalelayo yi-1K.
• size: Yamukela i-auto, 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, noma 21:9.
• n: Yamukela inombolo ephelele ukusuka ku-1 kuya ku-4.
• image_urls: Uhlu olungakhethwa oluqukethe ama-URL noma ama-data URL ayisethenjwa afinyelela kwangu-16.

Ukushayela umzila okhokhiswa ngamathokheni

Umzila osemthethweni usebenzisa indlela efanayo yokukhiqiza kodwa ushintsha i-ID yemodeli, ususe i-version, bese wengeza isilungiselelo sekhwalithi. Lesi sibonelo sikhetha i-Sunburst enekhwalithi ephezulu:

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
  }'

Bheka ireferensi yomzila osemthethweni kanye ireferensi yomzila odlulisiwe ukuze uthole izincazelo zamanje zezinkambu.

Amathiphu athuthukile okuhlanganisa okuthembekile

1. Yenza ukuzama futhi kube yi-idempotent. Gcina ukhiye wokuqala we-idempotency uma ukuthumela kuphelelwa isikhathi futhi ungaqiniseki ukuthi umsebenzi udaliwe yini. Ukhiye omusha ungadala okunye ukukhiqizwa okukhokhiswayo.
2. Thatha i-polling njengengxenye evamile yokusebenza. Phatha kokubili completed kanye failed. Gwema i-polling esheshayo; isibonelo esikurepository silinda imizuzwana emihlanu phakathi kokuhlola.
3. Sebenzisa i-URL ye-poll ebuyisiwe. Impendulo ihlanganisa data.poll_url, okusiza ugweme ukwakha kabusha i-endpoint yomsebenzi ngesandla.
4. Landela imiphumela ngokushesha. Ama-URL ezithombe ezikhiqiziwe awahlali unomphela. Kopisha umphumela ngamunye uwubeke endaweni yakho yokugcina ngaphambi kwesikhathi sawo se-expires_at.
5. Qopha izindleko zangempela. Gcina cost kanye credits_cost ezivela empendulweni yomsebenzi. Lokhu kukuvumela uqhathanise ukukhokhiswa okubonwe namanani ashicilelwe.
6. Qinisekisa izindleko ngesicelo esisodwa esikhokhelwayo. Amanani kanye nemizila yokukhushulwa kungashintsha, ngakho yenza ukuhlola okuncane ngaphambi kokuqalisa iqoqo elikhulu.
7. Sebenzisa umzila ofanele. Khetha umzila odlulisiwe ukuze uthole ukukhokhiswa okubikezeleka ngesithombe ngasinye namaqoqo afinyelela kwamane. Khetha i-ID yemodeli esemthethweni lapho izinga lekhwalithi noma ukulawula amaphikseli aqondile kubalulekile.
8. Shintsha izinhlobo ngaphandle kokubhala kabusha iklayenti. Emzileni odlulisiwe, shintsha kuphela version phakathi kuka-flare no-sunburst.
9. Vikela imininingwane yokungena. Gcina ukhiye we-API eziguquguqukayo zemvelo noma kumphathi wezimfihlo, futhi ungawuvezi nhlobo kukhodi esebenza esipheqululini.

Isiphetho

Isango le-APIMart le-image2.5 linikeza indlela engaguquki yokufinyelela ekukhiqizweni nasekuhleleni kwe-GPT Image 2.5. Qala nge-gpt-image-2.5-ext kanye ne-Flare ukuze uthole ukukhiqizwa okuningi, okubikezeleka, shintshela ku-Sunburst ukuze uhlele noma usebenze ngezithenjwa, futhi usebenzise ama-ID amamodeli asemthethweni lapho ukukhokhiswa ngamathokheni nezilawuli zekhwalithi ezengeziwe kuhambisana kangcono nomsebenzi.

Indlela ebalulekile yokuhlanganisa ilula: thumela isicelo sesithombe, gcina i-ID yomsebenzi ebuyisiwe, yenza i-poll kuze kuqedwe, hlola izindleko ezibikiwe, bese ulanda yonke imiphumela ngaphambi kokuphelelwa yisikhathi kwe-URL yayo yesikhashana. Ngaphambi kokukhulisa isikali, buyekeza amanani amanje e-APIMart futhi uqinisekise umzila owukhethile ngesicelo esincane esikhokhelwayo.