Peke ki ngā ihirangi
Āheinga

Āheinga

Te mea e taea e ia model mā te API, model ki te model: tools, input whakaahua, reasoning, output JSON, prompt caching, rapu tukutuku me te streaming.

Ka tae atu ki ia model mā ngā endpoint e toru ōrite me te key ōrite. Ka whakaatu ngā ripanga e rua i te wāhi e rerekē ai ngā model. Ko te tohuhono te tohu e whakautu ana te model kāore he taua āheinga.

Ngā model Shannon

Model ToolsWhakaahuaReasoningEffortJSON objectJSON schemaCachingRapuStreaming
shannon-1.6-lite ĀeĀe——ĀeĀe—ĀeĀe
shannon-1.6-pro ĀeĀeĀe—ĀeĀe—ĀeĀe
shannon-2-lite Āe—Āe—ĀeĀe—ĀeĀe
shannon-2-pro Āe—Āe—ĀeĀe—ĀeĀe
shannon-3 ĀeĀeĀe—ĀeĀe—ĀeĀe
shannon-3-pro ĀeĀeĀe—ĀeĀe—ĀeĀe
shannon-3.1 ĀeĀeĀe—ĀeĀe—ĀeĀe
shannon-3.1-pro ĀeĀeĀe—ĀeĀe—ĀeĀe
shannon-3.5 Āe———ĀeĀe——Āe
shannon-coder-1 Āe—Āe—ĀeĀe——Āe

Ngā model open-weight hosted

Ko te id o tētahi model hosted ko tōna ingoa me tōna momo kua honoa ki tētahi tohuhono, hei tauira DeepSeek-V4-Pro-0813-3BIT-REAP.

Model ToolsWhakaahuaReasoningEffortJSON objectJSON schemaCachingRapuStreaming
DeepSeek-V4-Pro-0813 3BIT-REAP Āe—ĀeĀeĀeĀeĀe—Āe
GLM-5.2 3BIT-REAP Āe—ĀeĀeĀeĀeĀe—Āe
Kimi-K3 3BIT-REAP ĀeĀeĀeĀeĀeĀeĀe—Āe
Nemotron3Ultra 3BIT-REAP Āe—ĀeĀeĀeĀeĀe—Āe
MiniMax-M3 3BIT-REAP ĀeĀeĀeĀeĀeĀeĀe—Āe
DeepSeek-V4-Flash-0731 W4A16-AUTOROUND-REAP Āe—ĀeĀeĀeĀeĀe—Āe
Kimi-K2.6 W4A16-AUTOROUND-REAP ĀeĀeĀeĀeĀeĀeĀe—Āe
Laguna-S-2.1 W4A16-AUTOROUND-REAP Āe—ĀeĀe——Āe—Āe
inkling W4A16-AUTOROUND-REAP ĀeĀeĀeĀeĀe—Āe—Āe
MiMo-V2.5-Pro W8A16 Āe—ĀeĀeĀeĀeĀe—Āe
MiMo-V2.5 W8A16 ĀeĀeĀeĀeĀeĀeĀe—Āe
Hy3 W8A16 Āe—ĀeĀeĀeĀeĀe—Āe

Te tikanga o ia pou

  • Tools. Ka whakaahua koe i ngā function i te tools, ka whakautu te model me ngā karanga e hiahia ana ia kia mahia, ā, ka whakahaere tō code i a rātou. Ka whakaae ia model ki ngā tools. Ka whakamahia e ngā model open-weight hosted tētahi tool_choice kua tohea. Karanga ā‑mahi
  • Input whakaahua. Ka pānui te model i ngā whakaahua i tukuna ki roto i tētahi karere kaiwhakamahi, hei data URL, hei wāhitau rānei. Ka pānui te shannon-2-lite me te shannon-2-pro i ngā whakaahua i ngā tono e tuku ana hoki i ngā tools, i te response_format rānei. Ka pānui hoki te whānau Shannon 3 i ngā tuhinga. Ngā whakaahua me ngā kōnae
  • Reasoning. Ka taea e te whakautu te kawe i te kuputuhi reasoning a te model i te taha o te whakautu, i tētahi āpure motuhake. Reasoning effort
  • Reasoning effort. Ka tautuhi tētahi āpure tono i te hōhonu o te reasoning a te model: off, low, medium he high rānei. Ka tautuhi ngā model Shannon i tō rātou ake hōhonutanga. Reasoning effort
  • JSON object. Me te response_format: {"type": "json_object"} he JSON te whakautu kāore he āhua kua whakaritea. I runga i te /v1/chat/completions me te /v1/responses. Putanga hanganga
  • JSON schema. Me te response_format o te momo json_schema ka whai te whakautu i tētahi schema e tukuna ana e koe. I runga i te /v1/chat/completions me te /v1/responses. Putanga hanganga
  • Prompt caching. Ka pānuitia mai i te cache tētahi prompt e tīmata ana pērā i tētahi o nā noa nei i runga i te model ōrite, ā, ka utua te wāhanga cached ki te 25% o te utu input. Kāore e hiahia ki tētahi āpure tono. Tūhono caching
  • Rapu tukutuku. Me te web_search: true ka rapu te API i te tukutuku ka hoatu ngā hua ki te model i mua i tana whakautu. I runga i te /v1/chat/completions me te /v1/messages. He āpure o te tono, ā, kāore he haki i te GET /v1/models. Rapu paetukutuku
  • Streaming. Me te stream: true ka tae mai te whakautu hei server-sent events i te wā e tuhi ana te model. Ka stream ia model i runga i ngā endpoint e toru katoa. Rere

Ngā meka ōrite mai i te GET /v1/models

GET https://api.shannon-ai.com/v1/models

Ka rārangi te GET /v1/models i ia model me tētahi capabilities object, kia taea e tētahi hōtaka te pānui i ēnei meka hei utu mō tēnei whārangi. Kāore te endpoint e hiahia ki te key.

from openai import OpenAI

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

for model in client.models.list():
    print(model.id, getattr(model, "capabilities", None))

Te id me te capabilities o ngā urunga e rua o te data:

200 JSON
[
  {
    "id": "shannon-3",
    "capabilities": {
      "tools": true,
      "vision": true,
      "reasoning": true,
      "reasoning_effort": false,
      "streaming": true,
      "json_schema": true
    }
  },
  {
    "id": "Kimi-K3-3BIT-REAP",
    "capabilities": {
      "tools": true,
      "vision": true,
      "reasoning": true,
      "reasoning_effort": true,
      "streaming": true,
      "response_format": true,
      "json_schema": true,
      "prompt_caching": true
    }
  }
]
Pou Haki i capabilities
Tools tools
Input whakaahua vision
Reasoning reasoning
Reasoning effort reasoning_effort
JSON object response_format
JSON schema json_schema
Prompt caching prompt_caching
Rapu tukutuku Ehara i te haki: te āpure tono web_search.
Streaming streaming
  • Ko te response_format me te prompt_caching he haki o ngā urunga open-weight hosted. Ka pūrongo tētahi urunga Shannon i te output JSON mā te json_schema anake, e tohu ana i ngā aratau JSON e rua.
  • Ko te vision he true mō shannon-2-lite me shannon-2-pro: ka pānui rāua i ngā whakaahua i ngā tono e tuku ana hoki i ngā tools, i te response_format rānei.

Kei waho i te rārangi o tētahi model

Ka whakakāhore tētahi model open-weight hosted i tētahi whakaahua, i tētahi aratau JSON rānei kei waho i tōna rārangi. Ka puta te whakakāhoretanga i mua i te tangohanga o tētahi mea i tō toenga:

Tūnga Momo Karere Āhea
400 invalid_request_error <id> does not accept image input I tukuna he whakaahua ki tētahi model open-weight hosted kāore he input whakaahua.
400 invalid_request_error <id> does not accept response_format I tonoa he output JSON i tētahi model open-weight hosted kāore he aratau JSON e rua.
400 invalid_request_error <id> does not accept response_format json_schema; use json_object I tukuna he schema ki tētahi model open-weight hosted he aratau JSON object anake.

Ka whakautu tētahi model Shannon ki taua tono me te tūnga 200: kāore te āheinga kei waho i tōna rārangi e uru ki te whakautu. Whakahaere hapa