Rārangi panoni

Ka ingoatia e tētahi tono me te web_search: true ōna pūtake: ka rārangitia e te sources i runga i te /v1/chat/completions me te /v1/messages ngā hua i hoatu ki te model, ā, i runga i te /v1/chat/completions ka tohu te annotations i hea e whakahua ai te whakautu i ia tētahi. Rapu paetukutuku

Ka whakahaere te web_search: true i runga i te /v1/chat/completions me te /v1/messages i tētahi rapu tukutuku i mua i te whakautu a te model, ā, ka hoatu ngā hua ki a ia hei puna kua tau. Ka whakamahi tētahi tono i kitea ai ngā hua i tētahi rapu o tō whakaaetanga i ia rā. Ina kore he toe, ka whakautua te tono kāore he rapu. Rapu paetukutuku

Ka utua tētahi stream ka tapahia i te taha server mō te usage i pūrongotia tae noa ki taua wā. Ka hoki te toenga o te rāhui ki tō toenga.

Ka whakautu te shannon-coder-1 i runga i te /v1/responses pērā i tāna i runga i ērā atu endpoint e rua. Ngā taputapu CLI mō te tuhi code

Ka mutu tonu te mahi a tētahi API key kua hurihia.

POST /v1/tokenize me POST /v1/messages/count_tokens: ngā tatau token kore utu mō ngā model open-weight hosted. Te tatau token

Prompt caching i runga i ngā model open-weight hosted. Ka utua te input cached ki te 25% o te utu input, ā, ka pūrongotia i te usage i runga i ngā endpoint e toru: prompt_tokens_details.cached_tokens, input_tokens_details.cached_tokens me cache_read_input_tokens. Ka pūrongo te GET /v1/models i te pricing.cached_input_per_million_usd me te capabilities.prompt_caching mō ēnei model. Tūhono caching

Ngā model open-weight hosted: ka puta he tool call i tētahi tool_choice kua tohea (required he function kua ingoatia rānei). Karanga ā‑mahi

Ka whakaae ngā model open-weight hosted ki tētahi karere system ehara i te karere tuatahi o te kōrero.

Ka taea te tae atu ki te API hei Tor onion service, ā, i tētahi wāhitau I2P. Ka kawe ngā whakautu o api.shannon-ai.com i ngā wāhitau e rua i ngā header Onion-Location me X-I2P-Location. Ngā endpoint whakamuna, Tor & I2P

Ngā model hōu shannon-3.1 me shannon-3.1-pro, kua rārangitia e te GET /v1/models. Ngā model me ngā utu

Ka rere tētahi tono stream ka motuhia tana kiritaki ki tōna mutunga, ā, ka tatauria. I mua, ka mutu te whakautu i te motunga.

Ka puta ngā tono ki te /v1/responses i te usage log.

Ka whakamahi ngā model open-weight hosted i ngā āpure sampling: temperature, top_p, seed, stop me max_tokens i runga i te /v1/chat/completions; temperature, top_p me stop_sequences i runga i te /v1/messages; temperature, top_p me max_output_tokens i runga i te /v1/responses. Chat Completions

Ka whakaatu te usage log i te utu USD mō ia karanga, me te tapeke.

Ka whakautua tētahi uara model ehara i te id kua whakaputaina ki te 400, te momo invalid_request_error me te karere unknown model: <id> i runga i ngā endpoint e toru. I mua, ka whakautua taua tono e te shannon-1.6-lite. Ngā model me ngā utu

Ka tango te /v1 i tētahi API key anake. Kāore te cookie takiuru o te pae tukutuku e pānuitia ki reira. Whakamotuhēhē

Flood protection: 120 tono i ia meneti mō ia pūkete i runga i ngā endpoint whakaputa e toru. Ngā tepe me te toenga

Ina kore e taea e te model te whakautu, kei te hapa te karere pūmau The model backend failed to answer. Please retry. Ka whakautua tētahi tirotiro toenga kāore e taea te mahi ki te 503 me te momo api_error. Whakahaere hapa

Ka whakahoki tētahi tono kāore i stream ka rahua i tōna rāhui ki tō toenga.

Ka pūrongo te GET /v1/models i te capabilities.reasoning_effort mō ia model. Reasoning effort

Ngā model open-weight hosted: ka mutu tētahi tono kāore i stream kāore e whiwhi whakautu me te 504 me te momo api_error. Ko te hua kau he 502. Whakahaere hapa

Ka tango ngā model open-weight hosted i te input e tangohia ana e te model kua ingoatia. Ka whakautua tētahi whakaahua i tukuna ki tētahi id kuputuhi-anake ki te 400 me te <id> does not accept image input. Ka whakautua te response_format i runga i tētahi id kāore he output whakatakoto ki te 400. Ka pūrongo te GET /v1/models i te capabilities.vision, response_format me json_schema mō ia id. Ngā whakaahua me ngā kōnae

Ka kawe ngā whakautu kāore i stream i te kuputuhi reasoning: reasoning_content i runga i te karere o te /v1/chat/completions, me tētahi poraka thinking i mua i te kuputuhi i runga i te /v1/messages. Reasoning effort

He utu hōu ō ngā model open-weight hosted, mai i te $0.50 mō ia 1M input tokens me te $2.00 mō ia 1M output tokens. Ngā model me ngā utu

Kotahi te id mō ia model open-weight hosted, tekau mā rua ngā id katoa. Ngā model me ngā utu

Output whakatakoto i runga i te shannon-3, shannon-3-pro, shannon-3.1 me shannon-3.1-pro. Ka pūrongo te GET /v1/models i te capabilities.json_schema: true mō rātou. Putanga hanganga

Ka pānui te /v1/responses i te text.format (json_object he json_schema rānei) mō te output whakatakoto. Responses API

Ka whakahoki te /v1/responses i te reasoning o ngā model Shannon: he item output reasoning i mua i te item message, me ngā tuhinga stream response.reasoning_summary_text.delta me response.reasoning_summary_text.done. Responses API

Ko te āpure model o tētahi whakautu te id kua whakaputaina i ia endpoint. Ka kawe tētahi Responses object i te model me te created_at.

Ka pūrongo te GET /v1/models i te matapihi horopaki o ia model. Ngā model me ngā utu

Ka whakautu ngā model open-weight hosted i runga i te /v1/messages me te /v1/responses i te taha o te /v1/chat/completions, me tō system prompt me ngā tool call. Ka tango te /v1/messages i te thinking, ā, ko te reasoning_effort he āpure tono. Ngā model me ngā utu

Ka whai ngā tinana hapa i runga i te /v1 i ngā āhua SDK: he error object me te type me te message i runga i ngā endpoint hōputu OpenAI, ā, te object ōrite i roto i te {"type": "error"} i runga i te /v1/messages. Whakahaere hapa

Keep-alive i runga i ngā stream tuwhera ia 15 hekona: he rārangi kōrero : ping i runga i te /v1/chat/completions, he event: ping i runga i te /v1/messages. Rere

Ka tīmata ngā id tool call ki te call_ i runga i te /v1/chat/completions me te /v1/responses, ā, ki te toolu_ i runga i te /v1/messages. Karanga ā‑mahi

/v1/responses: ko te taunoa o te stream ko false. Responses API

Ka whakaaetia te max_completion_tokens hei ingoa kē mō te max_tokens i runga i te /v1/chat/completions. Chat Completions

Ka kawe te tuhinga response.completed o tētahi stream Responses i te rārangi output. Responses API

Ka kawe ngā urunga GET /v1/models i te display_name, context_window, max_output_tokens, pricing, capabilities me endpoints. Ngā model me ngā utu

Ka kawe ia whakautu i te x-request-id, ā, ka whakahokia te uara e tukuna ana e koe. Ka taea e ngā tinana tono te tae ki te 32 MiB.

Ka taea te whakakore i tētahi key me te kore e tukuna he key hōu. Ka wehea ki ngā whārangi te usage log.

Ngā whakaahua me ngā tuhinga i ngā tono: ngā wāhanga image_url i runga i te /v1/chat/completions, ngā poraka image, document me file i runga i te /v1/messages, ngā wāhanga input_image i runga i te /v1/responses. Ngā whakaahua me ngā kōnae

Model hōu shannon-3-pro. Ngā model me ngā utu

Model hōu shannon-3. Ngā model me ngā utu

Ka kawe ngā stream i runga i te /v1/chat/completions i te reasoning hei delta.reasoning_content. Reasoning effort

Ka puritia te max_tokens ki roto i te 1 ki te 65,536. Ka kawea tētahi uara i waho i taua whānui ki te pito tata rawa, kāore e whakakāhoretia. Chat Completions

Ka tatauria te shannon-coder-1 i ngā karanga mō ia mahere i runga i te /v1/chat/completions me te /v1/messages. Ina pau ngā karanga o tētahi matapihi, ko te whakautu he 429 me te karere Shannon Coder call quota reached. Ngā taputapu CLI mō te tuhi code

Ka whakautua tētahi kōrero e nui ake ana i te matapihi horopaki ki te 400, te momo invalid_request_error me te waehere context_length_exceeded. Ka tae mai tētahi hapa i roto i tētahi stream hei tuhinga hapa, kaua hei kuputuhi i te whakautu. Ngā taputapu CLI mō te tuhi code

Ngā endpoint /v1: POST /v1/chat/completions, POST /v1/messages, POST /v1/responses me GET /v1/models. Koinei ngā endpoint e karangahia ana e ngā SDK OpenAI me Anthropic, Claude Code me Codex CLI. Tirohanga whānui

Ngā tool call me ngā tools i runga i te /v1/chat/completions me te /v1/messages, te output whakatakoto me te response_format, me te system hei rārangi tuhi, hei poraka kuputuhi rānei i runga i te /v1/messages. Karanga ā‑mahi

Ngā model id shannon-1.6-pro, shannon-2-lite, shannon-2-pro me shannon-coder-1, i te taha o te shannon-1.6-lite. Ngā model me ngā utu

Te tukunga tuatahi o te API: te chat i te hōputu OpenAI me te hōputu Anthropic, me tētahi endpoint Responses, ia tētahi ka karangahia mā tētahi API key e tīmata ana ki te sk-. Whakamotuhēhē