Tsallake zuwa abun ciki
Hotuna da fayiloli

Hotuna da fayiloli

Ku aika hotuna tare da saƙo: waɗanne models ke karɓa, siffofin da ake karɓa da iyakokinsu.

POST https://api.shannon-ai.com/v1/chat/completions

Hoto yana tafiya a cikin saƙon mai amfani, a matsayin sashe ɗaya na content ɗinsa. Samfurin yana karanta fayil na gida kuma yana aika shi a matsayin data URL.

import base64
from openai import OpenAI

client = OpenAI(api_key="YOUR_API_KEY", base_url="https://api.shannon-ai.com/v1")

with open("photo.jpg", "rb") as f:
    image = base64.b64encode(f.read()).decode()

response = client.chat.completions.create(
    model="shannon-3",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What is in this picture?"},
            {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64," + image}},
        ],
    }],
)

print(response.choices[0].message.content)
200 JSON
{
  "id": "chatcmpl-8d1f3a5c7e9b4d2f6a8c0e2b4d6f8a1c",
  "object": "chat.completion",
  "created": 1791590400,
  "model": "shannon-3",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "A grey cat asleep on a windowsill, with a potted fern beside it.",
        "reasoning_content": null
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 286,
    "completion_tokens": 19,
    "total_tokens": 305
  }
}

Waɗanne models ke karɓar hotuna

Models Yana karantawa
shannon-3, shannon-3-pro, shannon-3.1, shannon-3.1-pro Hotuna, da takardu: PDF, Word, PowerPoint, Excel da fayilolin rubutu kawai.
shannon-1.6-lite, shannon-1.6-pro Hotuna.
shannon-2-lite, shannon-2-pro Hotuna, a requests da kuma suke aika tools ko response_format.
Kimi-K3-3BIT-REAP, MiniMax-M3-3BIT-REAP, Kimi-K2.6-W4A16-AUTOROUND-REAP, inkling-W4A16-AUTOROUND-REAP, MiMo-V2.5-W8A16 Hotuna.
DeepSeek-V4-Pro-0813-3BIT-REAP, GLM-5.2-3BIT-REAP, Nemotron3Ultra-3BIT-REAP, DeepSeek-V4-Flash-0731-W4A16-AUTOROUND-REAP, Laguna-S-2.1-W4A16-AUTOROUND-REAP, MiMo-V2.5-Pro-W8A16, Hy3-W8A16 Rubutu. Ana amsa request mai hoto da matsayi 400.
shannon-coder-1 Rubutu.

GET /v1/models yana bayyana image input ga kowane id a matsayin capabilities.vision. Models & farashi

  • Models na Shannon suna karanta fayilolin saƙon mai amfani na ƙarshe. Ku sa hoton a saƙon da ke tambaya game da shi.
  • Iyalin Shannon 3 yana kuma riƙe wasu hotuna da takardu na saƙonnin mai amfani na baya a gani idan an aika su a cikin layi, kuma yana karanta iyakataccen matsakaicin adadin hotuna ga kowace request, ana ƙirga daga na farko.
  • Hosted open-weight models suna karanta hotunan kowane saƙo na tattaunawar.

Siffofin da ake karɓa

Ana aika fayil ta hanya ɗaya daga biyu: a cikin request a matsayin base64, ko a matsayin adireshin da API ke debowa. Kowane fayil yana tafiya tare da request; API ba shi da endpoint na upload.

Endpoint A cikin request (base64) Ta adireshi
/v1/chat/completions {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64,…"}} {"type": "image_url", "image_url": {"url": "https://…"}}
/v1/responses {"type": "input_image", "image_url": "data:image/jpeg;base64,…"} {"type": "input_image", "image_url": "https://…"}
/v1/messages {"type": "image", "source": {"type": "base64", "media_type": "image/jpeg", "data": "…"}} {"type": "image", "source": {"type": "url", "url": "https://…"}}

Saƙon mai amfani iri ɗaya a kowane tsari:

{
  "role": "user",
  "content": [
    {"type": "text", "text": "What is in this picture?"},
    {
      "type": "image_url",
      "image_url": {
        "url": "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD…"
      }
    }
  ]
}
  • Data URL yana da siffar data:<media type>;base64,<data>. Sashen ;base64 wajibi ne.
  • A /v1/chat/completions da /v1/responses, image_url na iya zama string ɗin da kansa ko object {"url": "…"}.
  • Nau'in fayil yana fitowa daga data URL ko daga media_type. Idan babu, ana karanta hoto a matsayin image/png.
  • Fayiloli na cikin layi suna cikin girman jikin request, wanda zai iya kaiwa 32 MiB.
  • Ana karɓa don jituwa: image_url.detail.

Iyakokin fayiloli na nesa

Idan kuka aika adireshi, API yana debo fayil kafin model ya gudana, a cikin waɗannan iyakoki:

Iyaka Ƙa'ida
Girma Har zuwa 8 MiB ga kowane fayil.
Lokaci Daƙiƙa 20 don amsawa.
Redirects Mafi yawa 5.
Adireshi http:// ko https://, ba tare da sunan mai amfani ko kalmar sirri a adireshin ba, a kan host mai adireshin jama'a.
Amsar host Matsayi a kewayon 200 da jiki wanda ba mara komai ba. Ana karanta nau'in fayil daga Content-Type.
Ana nema a matsayin User-Agent: ShannonBot/1.0 (+https://shannon-ai.com)

Yadda ake ƙirga hotuna a tokens

A hosted open-weight models ana ƙirga hoto token ɗaya ga kowane patch na pixels 28 × 28: faɗi da aka raba da 28, an zagaye sama, ninki tsayi da aka raba da 28, an zagaye sama. Sakamakon yana cikin usage.prompt_tokens kuma ana cajin sa a matsayin input.

1024 × 1024
1,369 tokens
1920 × 1080
2,691 tokens
512 × 512
361 tokens
  • Ana debo hoto na nesa da farko, don haka ana ƙirga ainihin girmansa.
  • Hoto da ba za a iya karanta girmansa ba ana ƙirga shi tokens 1,024.
  • Endpoints na ƙirga suna amfani da ƙa'ida iri ɗaya ga data URLs. Ba sa debo adireshi na nesa kuma suna ƙirga shi a matsayin tokens 1,024. Ƙidayar tokens
  • Ga model na Shannon, ku karanta abin da request mai hotuna ya kashe daga usage a cikin amsa.

Takardu da sauran fayiloli

Iyalin Shannon 3 (shannon-3, shannon-3-pro, shannon-3.1, shannon-3.1-pro) yana karanta takardu ma. Ana cire rubutun takardar a ba model tare da saƙonku.

Takarda Media type
PDF application/pdf
Word application/vnd.openxmlformats-officedocument.wordprocessingml.document
PowerPoint application/vnd.openxmlformats-officedocument.presentationml.presentation
Excel application/vnd.openxmlformats-officedocument.spreadsheetml.sheet

Takarda sashe ne na saƙon mai amfani, kamar hoto:

Endpoint A cikin request (base64) Ta adireshi
/v1/chat/completions, /v1/messages {"type": "document", "source": {"type": "base64", "media_type": "application/pdf", "data": "…"}} {"type": "document", "source": {"type": "url", "url": "https://…"}}
/v1/responses {"type": "input_file", "file_data": "data:application/pdf;base64,…"} {"type": "input_file", "file_url": "https://…"}
{
  "model": "shannon-3",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "document",
          "source": {
            "type": "base64",
            "media_type": "application/pdf",
            "data": "JVBERi0xLjcKJeLjz9MK…"
          }
        },
        {"type": "text", "text": "Summarise this report in five points."}
      ]
    }
  ]
}
  • Ku ba da ainihin media type na fayil ɗin. Idan babu, ana karanta takarda a matsayin application/pdf.
  • Fayil na rubutu kawai, kamar CSV, ana karanta shi a matsayin rubutu kuma a haɗa shi haka.
  • Idan ba a iya karanta takarda ba, ko fayil ɗin na wani nau'i ne, ana gaya wa model kuma yana iya faɗin hakan a amsarsa. Matsayin har yanzu 200 ne.
  • A kowane sauran model ana bar sashen takarda daga abin da model ke karantawa, kuma ana amsa request daga sauran.
  • Takardar da aka aika ta adireshi ana debo ta bisa iyakokin fayiloli na nesa a sama.

Kurakurai

Matsayi Nau'i Saƙo Lokaci
400 invalid_request_error <id> does not accept image input An aika hoto zuwa hosted open-weight model wanda ke karɓar rubutu kaɗai. Ba a cire komai daga balance ɗinku ba.
400 invalid_request_error audio and video input are not supported by any hosted model An aika sashen audio ko bidiyo zuwa hosted open-weight model.
413 invalid_request_error Failed to buffer the request body: … Jikin request, tare da fayilolinsa na cikin layi, ya fi 32 MiB girma. Kuskuren yana ɗauke da code: "request_too_large".

Model na Shannon yana amsawa da matsayi 200 komai fayilolin da request ke ɗauke da su: sashen da ba ya karantawa ana bar shi. Kuskuren Gudanarwa