Ndepụta mgbanwe

Request nwere web_search: true na-akpọ isi mmalite ya aha: sources na /v1/chat/completions na /v1/messages na-depụta results e nyere model, na na /v1/chat/completions annotations na-egosi ebe azịza na-ekwu ọ bụla. Chọọ weebụ

web_search: true na /v1/chat/completions na /v1/messages na-eme web search tupu model aza ma nye ya nsonaazụ dị ka isi mmalite e debara nọmba. Request chọtara nsonaazụ na-eji otu nchọ nke allowance gị kwa ụbọchị. Ọ bụrụ na ọ dịghị fọdụrụ, a na-azaghachi request na-enweghị nchọ. Chọọ weebụ

A na-akwụ ụgwọ stream e bepụrụ n'akụkụ server maka usage akọrọ ruo n'oge ahụ. Ihe fọdụrụ n'ime ihe e debere na-alọghachi na balance gị.

shannon-coder-1 na-azaghachi na /v1/responses dị ka ọ na-eme n'endpoints abụọ ndị ọzọ. Ngwa CLI maka coding

API key agbanwere na-akwụsị ịrụ ọrụ ozugbo.

POST /v1/tokenize na POST /v1/messages/count_tokens: ọnụọgụ token n'efu maka hosted open-weight models. Ịgụ tokens

Prompt caching na hosted open-weight models. A na-akwụ ụgwọ cached input na 25% nke ọnụahịa input ma a na-akọ ya n'ime usage n'endpoints atọ niile: prompt_tokens_details.cached_tokens, input_tokens_details.cached_tokens na cache_read_input_tokens. GET /v1/models na-akọ pricing.cached_input_per_million_usd na capabilities.prompt_caching maka models ndị a. Caching prompt

Hosted open-weight models: tool_choice e manyere (required ma ọ bụ function akpọrọ aha) na-ewepụta oku tool. Oku ọrụ

Hosted open-weight models na-anabata ozi system nke abụghị ozi mbụ nke mkparịta ụka.

Enwere ike iru API dị ka Tor onion service na n'adreesị I2P. Azịza api.shannon-ai.com na-ebu adreesị abụọ ahụ n'ime headers Onion-Location na X-I2P-Location. Encrypted, Tor & I2P endpoints

Models ọhụrụ shannon-3.1 na shannon-3.1-pro, GET /v1/models depụtara. Models & ọnụahịa

Request streamed nke client ya kwụsịrị njikọ na-agba ruo ọgwụgwụ ya ma a na-agụ ya. Na mbụ, azịza na-akwụsị mgbe njikọ kwụsịrị.

Requests ndị gaa /v1/responses na-apụta n'ime usage log.

Hosted open-weight models na-etinye sampling fields: temperature, top_p, seed, stop na max_tokens na /v1/chat/completions; temperature, top_p na stop_sequences na /v1/messages; temperature, top_p na max_output_tokens na /v1/responses. Chat Completions

Usage log na-egosi ọnụahịa na USD maka oku ọ bụla na dị ka ngụkọta.

A na-azaghachi uru model nke abụghị id e bipụtara na 400, ụdị invalid_request_error na ozi unknown model: <id> n'endpoints atọ niile. Na mbụ, shannon-1.6-lite na-azaghachi request dị otú ahụ. Models & ọnụahịa

/v1 na-anabata API key naanị. A naghị agụ sign-in cookie nke weebụsaịtị n'ebe ahụ. Nyocha njirimara

Flood protection: requests 120 kwa nkeji kwa akaụntụ n'endpoints mmepụta atọ. Limits na balance

Mgbe model enweghị ike ịzaghachi, njehie na-ebu ozi kwụsiri ike The model backend failed to answer. Please retry. A na-azaghachi nlele balance enweghị ike ime na 503 na ụdị api_error. Njikwa njehie

Request na-abụghị streamed nke na-ada na-enyeghachi balance gị ihe e debere.

GET /v1/models na-akọ capabilities.reasoning_effort maka model ọ bụla. Reasoning effort

Hosted open-weight models: request na-abụghị streamed nke na-enweghị azịza na-agwụ na 504 na ụdị api_error. Nsonaazụ efu bụ 502. Njikwa njehie

Hosted open-weight models na-anabata input nke model akpọrọ aha na-anabata. A na-azaghachi foto e zigara na id text-only na 400 na <id> does not accept image input. A na-azaghachi response_format na id na-enweghị structured output na 400. GET /v1/models na-akọ capabilities.vision, response_format na json_schema maka id ọ bụla. Images na files

Azịza na-abụghị streamed na-ebu reasoning text: reasoning_content n'ozi nke /v1/chat/completions, na thinking block tupu text na /v1/messages. Reasoning effort

Hosted open-weight models nwere ọnụahịa ọhụrụ, site na $0.50 kwa 1M input tokens na $2.00 kwa 1M output tokens. Models & ọnụahịa

Hosted open-weight models nwere otu id kwa model, ids iri na abụọ n'ozuzu. Models & ọnụahịa

Structured output na shannon-3, shannon-3-pro, shannon-3.1 na shannon-3.1-pro. GET /v1/models na-akọ capabilities.json_schema: true maka ha. Mmepụta ahaziri

/v1/responses na-agụ text.format (json_object ma ọ bụ json_schema) maka structured output. Responses API

/v1/responses na-eweghachi reasoning nke models Shannon: reasoning output item tupu message item, na stream events response.reasoning_summary_text.delta na response.reasoning_summary_text.done. Responses API

Field model nke azịza bụ id e bipụtara n'endpoint ọ bụla. Responses object na-ebu model na created_at.

GET /v1/models na-akọ context window nke model ọ bụla. Models & ọnụahịa

Hosted open-weight models na-azaghachi na /v1/messages na /v1/responses na /v1/chat/completions nakwa, site na system prompt gị na oku tool. /v1/messages na-anabata thinking, ma reasoning_effort bụ request field. Models & ọnụahịa

Error bodies na /v1 na-eso udi SDK: error object nwere type na message n'endpoints n'usoro OpenAI, na otu object ahụ n'ime {"type": "error"} na /v1/messages. Njikwa njehie

Keep-alive na streams meghere kwa sekọnd 15: ahịrị nkọwa : ping na /v1/chat/completions, event: ping na /v1/messages. Nzipu ozugbo

Ids nke oku tool na-amalite na call_ na /v1/chat/completions na /v1/responses, ma na toolu_ na /v1/messages. Oku ọrụ

/v1/responses: stream default bụ false. Responses API

A na-anabata max_completion_tokens dị ka aha ọzọ maka max_tokens na /v1/chat/completions. Chat Completions

response.completed event nke Responses stream na-ebu output array. Responses API

Entries GET /v1/models na-ebu display_name, context_window, max_output_tokens, pricing, capabilities na endpoints. Models & ọnụahịa

Azịza ọ bụla na-ebu x-request-id, a na-ewetaghachikwa uru ị zipụrụ. Request bodies nwere ike ịbụ ruo 32 MiB.

Enwere ike ịkagbu key na-enweghị ịnye nke ọhụrụ. A na-ekewa usage log n'peeji.

Foto na akwụkwọ n'ime requests: akụkụ image_url na /v1/chat/completions, blocks image, document na file na /v1/messages, akụkụ input_image na /v1/responses. Images na files

Model ọhụrụ shannon-3-pro. Models & ọnụahịa

Model ọhụrụ shannon-3. Models & ọnụahịa

Streams na /v1/chat/completions na-ebu reasoning dị ka delta.reasoning_content. Reasoning effort

A na-edebe max_tokens n'ime 1 ruo 65,536. A na-ewetara uru dị n'èzí oke ahụ nso na ọnụ kacha nso, a naghị ajụ ya. Chat Completions

A na-agụ shannon-coder-1 n'ime oku kwa plan na /v1/chat/completions na /v1/messages. Mgbe oku nke windo gwụrụ, azịza bụ 429 nwere ozi Shannon Coder call quota reached. Ngwa CLI maka coding

A na-azaghachi mkparịta ụka karịrị context window na 400, ụdị invalid_request_error na koodu context_length_exceeded. Njehie n'ime stream na-abịa dị ka error event, ọ bụghị dị ka text n'ime azịza. Ngwa CLI maka coding

Endpoints /v1: POST /v1/chat/completions, POST /v1/messages, POST /v1/responses na GET /v1/models. Ha bụ endpoints SDK OpenAI na Anthropic, Claude Code na Codex CLI na-akpọ. Nchịkọta

Oku tool site na tools na /v1/chat/completions na /v1/messages, structured output site na response_format, na system dị ka string ma ọ bụ dị ka text blocks na /v1/messages. Oku ọrụ

Model ids shannon-1.6-pro, shannon-2-lite, shannon-2-pro na shannon-coder-1, n'akụkụ shannon-1.6-lite. Models & ọnụahịa

Mbipụta mbụ nke API: chat n'usoro OpenAI na n'usoro Anthropic, na Responses endpoint, nke ọ bụla a na-akpọ site na API key malitere na sk-. Nyocha njirimara