Dala Izithombe nge-GPT Image 2.5 Usebenzisa i-APIMart API

Lokho Okunikezwa Yile Phrojekthi

I-Image2.5 Prompt Gallery iqoqo elisebenzayo lama-prompt angu-12 asetshenziswe nge-GPT Image 2.5 futhi ashicilelwa kanye nezithombe eziphumayo. Iresiphi ngayinye irekhoda i-prompt, inguqulo yemodeli, isilinganiso sobude nobubanzi, kanye nomphumela, okwenza kube lula ukuhlola i-prompt ngaphambi kokuyilungisa ukuze uyisebenzise kuhlelo lwakho.

Izibonelo zisebenzisa i-endpoint ehambisana ne-OpenAI ye-APIMart ethi https://api.apimart.ai/v1. Umzila odluliselwe usebenza ngokungavumelani: uhlelo lwakho luthumela isicelo sesithombe, luthole i-ID yomsebenzi, luhlole i-endpoint yomsebenzi ngezikhathi ezithile, bese ludawuniloda isithombe ngemva kokuqedwa kokucutshungulwa.

Izici Eziyinhloko

• Indlela yokukhiqiza ehambisana ne-OpenAI: izicelo zithunyelwa ku-POST /v1/images/generations.
• Izinhlobo ezimbili ezidluliselwe: sebenzisa version: "flare" ekukhiqizeni kwansuku zonke noma version: "sunburst" ekuhleleni nasemsebenzini oqondiswa yizithenjwa.
• Amazinga amaningi okuphuma: khetha 1K, 2K, noma 4K.
• Ukwenza uhlaka ngendlela evumelana nezidingo: sebenzisa auto noma esinye sezilinganiso zobude nobubanzi eziyishumi ezisekelwayo, okuhlanganisa 1:1, 16:9, 9:16, kanye 4:3.
• Ukuphuma kwenqwaba: cela izithombe eziphakathi kwesisodwa nezine usebenzisa n.
• Ukusekelwa kwezithombe eziyizithenjwa: nikeza ama-URL noma ama-data URL afinyelela ku-16 ku-image_urls.
• Ukubikwa kwezindleko zesicelo ngasinye: idatha yomsebenzi oqediwe ihlanganisa cost kanye credits_cost.
• Ukuphinda izicelo ngaphandle kokuphinda umphumela: i-API yamukela i-header ethi Idempotency-Key ukusiza ukuvimbela umsebenzi ophindiwe lapho uthumela isicelo esifanayo futhi.

Khetha Umzila Wemodeli Ofanele

I-APIMart inikeza imizila ye-GPT Image 2.5 edluliselwe kanye nesemthethweni. Izinkambu zezicelo nezindlela zokukhokhisa zihlukile, ngakho khetha umzila ngaphambi kokubhala ukuhlanganisa kwakho.

Umzila odluliselwe okhokhiswa ngesithombe ngasinye

Setha model ku-gpt-image-2.5-ext bese ufaka version: "flare" noma version: "sunburst". Lona umzila osetshenziswa yizibonelo zegalari futhi ukhokhiswa ngesithombe ngasinye esilethiwe.

• flare: ulungele ukukhiqiza kwansuku zonke kanye nemidwebo yokuqala yenqwaba.
• sunburst: ulungele ukunemba ekuhleleni nasemsebenzini okufanele ulondoloze izithenjwa, izinto, amalogo, noma amalebula emikhiqizo.

Umzila osemthethweni okhokhiswa ngamathokheni

Sebenzisa gpt-image-2.5-flare noma gpt-image-2.5-sunburst ngokuqondile. Ungathumeli inkambu ethi version nalawa ma-ID amamodeli. Umzila osemthethweni ukhokhiswa ngamathokheni futhi usekela amazinga ekhwalithi asuka ku-low aye ku-max, kanye nobukhulu obuqondile ngamaphikseli.

Ungawahlanganisi amafomethi emizila. I-alias ethi gpt-image-2.5-ext idinga isikhethi senguqulo, kanti ama-ID amamodeli asemthethweni akumele ayifake.

Ukufaka Nokulungisa

1. Dala bese uxhasa ngemali i-akhawunti

Thola ukhiye we-API ekhasini lokhiye le-APIMart. Ngokwerepositori, isevisi ayinaso isabelo samahhala futhi isekela ukugcwalisa imali okuqala ku-$1.

2. Gcina ukhiye ku-variable yemvelo

export APIMART_API_KEY="your-api-key"

Gwema ukufaka ukhiye ngokuqondile kukhodi yomthombo yohlelo noma ukuwufaka ohlelweni lokulawula izinguqulo.

3. Faka i-Python dependency

Isibonelo se-Python sisebenzisa iphakheji yomuntu wesithathu ethi requests.

python -m pip install requests

Uma ukhetha i-curl, qinisekisa ukuthi i-curl kanye nejeneretha ye-UUID efana ne-uuidgen kuyatholakala ku-shell yakho.

Yenza Isicelo Sakho Sokuqala Sesithombe Nge-curl

Isicelo esilandelayo sakha isithombe esiyisikwele esingu-1K sisebenzisa uhlobo oludluliselwe lwe-flare:

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": "Indawo yokufunda ethokomele eduze kwefasitela ngosuku olunemvula, isibani setafula esifudumele, ukukhanya kwesinema",
    "size": "1:1",
    "resolution": "1K",
    "n": 1
  }'

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

Hlola umsebenzi

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

Umsebenzi ngokuvamile usuka ku-pending uye ku-processing, bese uqeda ngokuba completed noma failed. Lapho uqeda, thola ama-URL akhiqiziwe ku-result.images[].url.

Ama-URL ezithombe ezikhiqiziwe afaka isitembu sesikhathi esithi expires_at. Dawuniloda futhi ugcine ifayela ngalinye ngaphambi kokuphelelwa yisikhathi kwe-URL yalo.

Yakha Iklayenti Le-Python Eliphelele Kusukela Ekuqaleni Kuya Ekugcineni

Lesi sibonelo sithumela isicelo esisodwa, sihlola amaphutha e-HTTP, sihlola njalo ngemva kwemizuzwana emihlanu, bese sibuyisa irekhodi lomsebenzi oqediwe. Sakha ukhiye omusha we-idempotency ekukhiqizeni ngakunye okusha.

import os
import time
import uuid
import requests

BASE_URL = "https://api.apimart.ai/v1"
API_KEY = os.environ["APIMART_API_KEY"]

COMMON_HEADERS = {
    "Authorization": f"Bearer {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,
        timeout=60,
        json={
            "model": "gpt-image-2.5-ext",
            "version": version,
            "prompt": prompt,
            "size": size,
            "resolution": resolution,
            "n": 1,
        },
    )
    response.raise_for_status()
    task_id = response.json()["data"]["id"]

    while True:
        poll_response = requests.get(
            f"{BASE_URL}/tasks/{task_id}",
            headers=COMMON_HEADERS,
            timeout=60,
        )
        poll_response.raise_for_status()
        task = poll_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(
    "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"
)

print("Cost:", result.get("cost"))
for image in result["result"]["images"]:
    print("Image URL:", image["url"])
    print("Expires at:", image.get("expires_at"))

Zama ngokuphepha

Uma ukuthunyelwa kuphelelwa isikhathi noma kuthola impendulo engazanywa futhi, thumela kabusha isicelo esifanayo esinengqondo usebenzisa i-Idempotency-Key efanayo. Sebenzisa ukhiye omusha lapho udala isithombe esisha ngenhloso. Uma umsebenzi oqediwe kufanele wenziwe kabusha ngenxa yokuthi umthombo noma ireferensi ishintshile, sebenzisa nokhiye omusha.

Qonda Izinkambu Eziyinhloko Zesicelo

• model: iyadingeka; sebenzisa i-gpt-image-2.5-ext emzileni odluliswayo.
• version: flare noma sunburst; okuzenzakalelayo ngu-flare.
• prompt: iyadingeka futhi kufanele ibe nombhalo ongagcwali ngezikhala kuphela.
• resolution: 1K, 2K, noma 4K; okuzenzakalelayo ngu-1K.
• size: okuzenzakalelayo ngu-auto. Izilinganiso ezisekelwayo yi-1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, kanye ne-21:9.
• n: inani lezithombe, kusukela kwesisodwa kuye kwezine emzileni odluliswayo.
• image_urls: uhlu olungakhethwa oluqukethe ama-URL ezithombe noma ama-data URL afinyelela kwayi-16.

Bhala Ama-Prompt Anokwethenjelwa Kakhulu

Igalari iphatha i-prompt njengombhalo omfishane wokukhiqiza, hhayi njengohlu lwezichasiso ezingacacile. Isakhiwo esiwusizo sihlanganisa into egxilwe kuyo, impahla, ukukhanya, ukubiyela noma ilensi, kanye nendlela yokuqeda.

1. Yisho into egxilwe kuyo: chaza into eyinhloko, umuntu, indawo, noma imvelo.
2. Cacisa izinto ezibonakalayo: sebenzisa imishwana efana ne-matte ceramic, frosted glass, noma clay render.
3. Setha ukukhanya: chaza ukukhanya okuthambile kwefasitela, ukukhanya kwemvelo okuvela eceleni, ukukhanya kwesitudiyo okukhanyayo, noma ukukhanya okunamandla okuvela ngemuva.
4. Lungisa ukubiyela: chaza ukwakheka okugxile phakathi, ukubuka okuvela phezulu, ukwakheka okubanzi kwesinema, noma enye indlela yokuhlela oyihlosile.
5. Faka ulimi lwelensi: amagama afana ne-85mm, 24mm, macro, kanye ne-focus stacked angasiza ukuqondisa umbono nemininingwane.
6. Khetha indlela yokuqeda: imishwana efana nemininingwane efana nomdwebo, ukuthwebula izithombe zomkhiqizo kwezohwebo, noma imiphetho ye-vector ebukhali ilawula indlela yokubukeka.

Isibonelo:

Cosmetics packshot of a frosted glass bottle with amber serum,
water droplets on the surface, seamless pale pink backdrop,
high detail commercial product photography

Sebenzisa Izithombe Zereferensi

Ngomsebenzi wokuguqula isithombe sibe isithombe noma osebenzisa ireferensi, faka uhlu lwe-image_urls esicelweni esidluliswayo. Umzila wamukela amareferensi afinyelela kwayi-16.

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

Sebenzisa ama-URL ereferensi azinzile futhi afinyeleleka. Uma umsebenzi wehluleka ngenxa yokuthi i-URL yereferensi iphelelwe isikhathi, phinda ubambe isithombe kwenye indawo bese uthumela umsebenzi omusha ngokhiye omusha we-idempotency.

Shayela Umzila Osemthethweni

Umzila osemthethweni usebenzisa i-ID yemodeli eqondile kanye nesilungiselelo se-quality esikhundleni sesikhethi se-version somzila odluliswayo.

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 imibhalo yomzila osemthethweni kanye nemibhalo yomzila odluliswayo ukuze uthole imininingwane yamanje yezinkambu.

Amathiphu Athuthukile Okukhiqiza

Hlukanisa ukuthumela nokucubungula

Gcina i-ID yomsebenzi ebuyisiwe kulayini noma kusizindalwazi esikhundleni sokugcina isicelo sewebhu sivulekile. Isisebenzi sangemuva singahlola imisebenzi engakaqedi futhi sigcine imiphumela eseqediwe.

Sebenzisa izikhawu zokuhlola ezimaphakathi

Izibonelo zenqolobane zihlola njalo ngemizuzwana emihlanu. Gwema amaluphu aqinile adala ithrafikhi engadingekile. Emisebenzini emikhulu, engeza ukwandisa izikhawu kancane ngenkathi umsebenzi usalindile noma usacutshungulwa.

Qopha izindleko zangempela

Gcina i-cost noma i-credits_cost kuwo wonke umsebenzi oqediwe. Lokhu kukunikeza irekhodi lokusetshenziswa kwemali ngocingo ngalunye esikhundleni sokuthembela ezilinganisweni kuphela.

I-README ibika inani elibonwe lomzila odluliswayo elingu-$0.0085 ngesithombe esingu-1K esilethiwe, u-$0.014 ku-2K, kanye no-$0.021 ku-4K flare. Inani le-4K sunburst alizange linqunywe kuthebula. Lezi zibalo zaqoshwa ngoSepthemba 2026, ngakho hlola amanani abukhoma ngaphambi kokwenza isabelomali noma ukwandisa umthamo.

Landa imiphumela ngokushesha

Ungathathi i-result.images[].url njengendawo yokugcina engunaphakade. Landa ifayela ngaphambi kwe-expires_at, bese ulidlulisela endaweni yokugcina elawulwa uhlelo lwakho lokusebenza.

Qala kancane ngaphambi kokwandisa

Qinisekisa isikhuthazo ku-1K nge-n: 1. Khulisa ukucaca noma ucele imiphumela eminingi kuphela ngemva kokuba isihloko, ukwakheka, nesitayela sekuzinzile.

Qondanisa i-aspect ratio nendawo oya kuyo

• 1:1 yezithonjana, amakhadi emikhiqizo, nezimpahla zokuxhumana nabantu eziyisikwele.
• 16:9 yezigcawu zesinema, amadeshibhodi, nezihloko zezethulo.
• 9:16 yezakhiwo eziqondile zeselula nezokuxhumana nabantu.
• 4:3 yemifanekiso yokuhlela nezithombe zokudla.
• 3:4 yezithombe zobuso nokwakheka kwemfashini.

Ukuxazulula Izinkinga

• 401 noma ukhiye we-API ongavumelekile: kopisha ukhiye futhi bese uhlola ukuthi awunazo yini izikhala noma izinhlamvu zomugqa omusha. Isihloko kufanele sisebenzise i-Authorization: Bearer $APIMART_API_KEY.
• Inkinga yebhalansi noma yekhredithi: faka imali ku-akhawunti; inqolobane ithi ayikho i-quota yamahhala eyisipele.
• Impendulo engu-429: yehlisa inani lemisebenzi ehambisanayo, yandisa isikhawu, bese uzama futhi ukuthumela okufanayo usebenzisa ukhiye ofanayo we-idempotency.
• 400 noma imodeli ayitholakali: qinisekisa ukuthi i-gpt-image-2.5-ext ifaka inguqulo nokuthi ama-ID emodeli asemthethweni aqondile awanayo.
• Isimo somsebenzi sithi failed: hlola ukuthi isikhuthazo senqatshiwe yini noma i-URL yereferensi iphelelwe yisikhathi. Lungisa okufakwayo bese uthumela umsebenzi omusha ngokhiye omusha.
• I-URL yomphumela ayisasebenzi: kungenzeka ukuthi i-URL yesikhashana isiphelelwe yisikhathi. Imiphumela ezayo eseqediwe kufanele ilandwe ngokushesha.

Isiphetho

Igalari Yezikhuthazo ye-image2.5 ikhombisa inqubo ephindaphindekayo yokukhiqiza izimpahla nge-GPT Image 2.5: khetha umzila ofanele, bhala isikhuthazo sesitayela sokukhiqiza, thumela isicelo esingavumelanisiwe, hlola i-ID yomsebenzi, qopha izindleko ezibikiwe, bese ulanda umphumela ngaphambi kokuphelelwa yisikhathi. Qala ngesithombe esisodwa se-1K, uqinisekise ikhwalithi nokukhokhiswa, bese udlulela ekucaceni okuphezulu, kumaqoqo, noma ezinqubweni ze-sunburst eziqhutshwa ngamareferensi.