U gudub nuxurka
Xadad iyo haraag

Xadad iyo haraag

Codsi kasta si siman ayaa loo bixiyaa. Heerar rate ma jiraan. Quota API oo gooni ah ma jirto. Horeba token-yadaada ayaad bixisay — isticmaal si degdeg ah sida aad rabto.

Boggani wuxuu sharxayaa waxa haraaggaagu ka kooban yahay, waxa hal codsi keydiyo oo ku kacayo, imisa codsi ayaad diri kartaa, iyo xadadka yar ee hal codsi la kulmi karo.

qiimaha 1M token oo haraag ah
$5.00
allowance-ka maalinlaha ah wuu cusboonaadaa
00:00 UTC
flood protection, akoon kasta
120 codsi / daqiiqad

Sida codsiyada loo bixiyo

  • Heerar rate ma jiraan — Hal xeer ayaa xadidaa sida degdegga ah ee codsiyadu u imaan karaan, wuxuuna isku mid yahay akoon kasta iyo qorshe kasta: 120 codsi daqiiqad kasta. Xad token-yo daqiiqad kasta ma jiro.
  • Quota API oo gooni ah ma jirto — API-gu wuxuu isticmaalaa isla haraagga chat-ka. Qorshuhu wuxuu dejiyaa cabbirka allowance-ka maanta. Ma dejiyo heerka codsiyada.
  • Si degdeg ah sida aad rabto — Codsiyada isku mar la diro waa la aqbalaa oo safka ayay ku sugaan. Lagama diido inay isku mar yihiin.

Haraaggaaga

Haraaggaaga waxaa lagu tiriyaa token-yo. 1,000,000 token haraag ah waxay u qiimo leeyihiin $5.00, qiime kasta oo bogga Model-yo & qiimo ahna waa heer ku saabsan qiimahaas.

Wakhti kasta haraagga waa wadarta laba qaybood.

  • Allowance-ka qorshaha ee maanta — Tiro token-yo ah oo qorshahaagu dejiyo. Maalin kasta saacadda 00:00 UTC wuu cusub yahay. Waxa maalinta dhammaadkeeda hadha lama sii wado.
  • Credit la iibsaday — Token-yo aad ku iibsatay xirmo. Credit-ku ma dhacayo wuxuuna ka shaqeeyaa qorshe kasta, Free ku jira.
Qorshe Token-yo maalintii Qiimo
Free 30,000 $0.15
Plus 80,000 $0.40
Standard 265,000 $1.325
Pro 665,000 $3.325
  • Kala horreynta isticmaalka — Codsi kasta wuxuu marka hore isticmaalaa allowance-ka qorshaha ee maanta. Credit la iibsaday waxaa la isticmaalaa wixii allowance-ka maalintaas dhaafa oo keliya.
  • Chat-ka iyo API-gu way wadaagaan — Akoon kasta wuxuu leeyahay hal haraag. Furaha API wuxuu ka isticmaalaa haraagga akoonka leh, isla qiimayaasha chat-ka.
  • Xirmooyin — Credit waxaa laga iibiyaa xirmooyin 1,000,000 ($5.00), 2,000,000 ($10.00) iyo 5,000,000 ($25.00) token ah, ama cadad aad dooratid 1,000,000 ilaa 100,000,000 token oo $5.00 halkii 1,000,000.

Kordhi credit-ka Beddel qorshaha

Waxa codsi keydiyo iyo waxa ku kacayo

  • Keydi — Marka codsi yimaado wuxuu haraaggaaga ka keydiyaa miisaaniyadda output-kiisa: max_tokens /v1/chat/completions iyo /v1/messages, max_output_tokens /v1/responses. /v1/chat/completions wuxuu sidoo kale akhriyaa max_completion_tokens. Default-ku waa 4,096 kala duwanaanshuhuna waa 1 ilaa 65,536.
  • Aqbal — Codsiga waa la aqbalaa haddii keydku ku soo galo waxa ka hadhay haraaggaaga oo keliya. Haraag eber ka weyn laakiin ka yar miisaaniyadda output-ka wuxuu helaa jawaabta Quota exceeded. Dir max_tokens yar si aad u isticmaasho inta kale.
  • Xisaab-xir — Marka jawaabtu dhammaato, keydka waxaa lagu beddelaa dallaca runta ah. Dallacu wuu ka hooseyn karaa ama ka sarreyn karaa keydka.
  • Celi — Codsi ku dhammaada xaalad khalad wuxuu keydkiisa dhammaystiran ku celiyaa.

Dallaca runta ahi wuxuu ku xiran yahay qoyska model-ka.

Model-yo Waxa la dallaco
Model-yada Shannon usage.total_tokens qiimaha model-ka halkii 1M. Input iyo output waxay leeyihiin hal qiime.
Model-yada open-weight ee la martigeliyo Input aan cached ahayn qiimaha input-ka, input cached qiimaha cached, output qiimaha output-ka.

Cadadka USD ah waxaa laga qaadaa haraaggaaga token-yo oo $5.00 halkii 1,000,000, lagu wareejiyay token dhan.

Tirinta token-yada POST /v1/tokenize ama POST /v1/messages/count_tokens waa bilaash waxna ma keydiyaan. Tirinta token-yada

Halka laga arko haraagga iyo isticmaalka

Bogga Fure & isticmaal wuxuu muujiyaa waxa aad hadda isticmaali karto, allowance-ka qorshaha ee maanta, credit-kaaga la iibsaday iyo kharashka API ee 30-kii maalmood ee la soo dhaafay. Hoosta wuxuu liis garaynayaa codsi kasta oo furahaagu sameeyay: wakhti, endpoint, model, input cached, token-yada la dallacay iyo kharash. Fure & isticmaal

Jawaab kasta waxay sidoo kale sidaa shay usage ah oo leh tirooyinka token-yada wicitaankaas.

Endpoint Field-yada usage Ay ku daraan model-yada open-weight ee la martigeliyo
/v1/chat/completions prompt_tokens, completion_tokens, total_tokens prompt_tokens_details.cached_tokens, completion_tokens_details.reasoning_tokens
/v1/messages input_tokens, output_tokens cache_read_input_tokens, cache_creation_input_tokens
/v1/responses input_tokens, output_tokens, total_tokens input_tokens_details.cached_tokens, output_tokens_details.reasoning_tokens
  • usage wuxuu haystaa tirooyinka token-yada model-ka. Cadadka haraaggaaga laga qaaday jawaabta kuma jiro: waa tiirka Token-yada la dallacay ee liiska codsiyada ee Fure & isticmaal.
  • /v1/messages oo leh model open-weight oo la martigeliyay, input_tokens waa qaybta input-ka aan cached ahayn, cache_read_input_tokens waa qaybta cached, cache_creation_input_tokens-na had iyo jeer waa 0.
  • Stream /v1/chat/completions wuxuu usage ku sidaa chunk-kiisa ugu dambeeya ka hor [DONE]. Streaming

Marka haraagga dhammaado

Codsi keydkiisu ku soo galin haraaggaaga waxaa lagu jawaabaa xaalad 429, nooca rate_limit_error iyo fariinta hoose. Waxba lagu ma dallaco. Isla jawaabtaas ayaa la diraa marka haraagga eber ka weyn yahay laakiin ka yar yahay miisaaniyadda output-ka ee codsiga.

{
  "error": {
    "type": "rate_limit_error",
    "message": "Quota exceeded. Upgrade your plan at shannon-ai.com/plan"
  }
}

/v1/responses shayga error wuxuu sidoo kale haysan karaa code iyo param, labaduba null.

Waxa aad sameyn karto:

  • Sug allowance-ka qorshaha ee xiga saacadda 00:00 UTC.
  • Kordhi credit. Credit waxaa la isticmaalaa allowance-ka qorshaha ka dib wuuna dhacayo. Kordhi credit-ka
  • U beddel qorshe leh allowance maalinle ah oo ka weyn. Beddel qorshaha
  • Dir max_tokens yar, haddii haraag yar ka hadhay: keydku kaddib wuu yaraanayaa.

Allowance-ka wicitaannada Shannon Coder

shannon-coder-1 /v1/chat/completions iyo /v1/messages waxaa lagu tiriyaa wicitaanno, ma aha token-yo. Qorshe kastaa wuxuu ka kooban yahay tiro wicitaanno ah daaqad 4-saac ah. Hal codsi waa hal wicitaan.

Qorshe Wicitaanno daaqad 4-saac ah
Free 3
Plus 20
Standard 40
Pro 60
  • Daaqadaha waxay bilaabmaan 00:00, 04:00, 08:00, 12:00, 16:00, 20:00 UTC. Wicitaannada daaqadda dhammaadkeeda hadha lama sii wado.
  • Wicitaan waa la tiriyaa marka codsiga la aqbalo, ka hor inta model-ku jawaabin. Codsi ka dib fashilma weli wuxuu tirsan yahay wicitaan.
  • Wicitaannadani token ma keydiyaan haraaggaagana waxba ka qaadaan. Liiska codsiyada ee Fure & isticmaal wuxuu muujiyaa tirada token-yadooda iyo qiimahooda qiimaha la liis gareeyay.
  • max_tokens default-ka ee shannon-coder-1 labadan endpoint waa 65,536.
  • Wicitaanno la'aan jawaabtu waa xaalad 429, nooca rate_limit_error, fariinta Shannon Coder call quota reached. Upgrade your plan at shannon-ai.com/plan.
  • /v1/responses, shannon-coder-1 ma laha allowance wicitaanno: waxaa lagu dallacaa token-yo haraaggaaga oo $8.00 halkii 1M, sida model kasta oo kale.

Flood protection

Akoon wuxuu diri karaa 120 codsi daqiiqad kasta. Kani waa xadka keliya ee heerka codsiyada, wuxuuna isku mid yahay qorshe kasta. Wuxuu u jiraa in lagu joojiyo daadad, ma aha in isticmaalka caadiga ah la dhimo.

  • Daqiiqaddu waa daaqad go'an oo 60 ilbiriqsi ah oo ku furmata codsigaaga ugu horreeya. Marka ay dhammaato, tirintu mar kale waxay ka bilaabmaan eber.
  • Tirintu waa akoon kasta, ma aha fure kasta ama cinwaan IP kasta. Beddelka furuhu daaqad cusub ma furo.
  • Codsiga 121aad daaqad dhexdeeda waxaa lagu jawaabaa xaalad 429, nooca rate_limit_error iyo fariinta Too many requests. Retry in <N>s. N waa tirada ilbiriqsiyada ilaa daaqaddu dhammaato, 1 ilaa 60.
  • Flood protection waa la hubiyaa ka hor haraagga. Codsi uu diido waxba ma keydiyo wax kumana kacayo.
{
  "error": {
    "type": "rate_limit_error",
    "message": "Too many requests. Retry in 37s."
  }
}
Codsi Flood protection
POST /v1/chat/completions, POST /v1/messages, POST /v1/responses Waa la tiriyaa, hal codsi kasta.
GET /v1/models, POST /v1/tokenize, POST /v1/messages/count_tokens Lama tiriyo.
shannon-coder-1 /v1/chat/completions iyo /v1/messages Waxaa lagu tiriyaa allowance-ka wicitaannada Shannon Coder halkii.
Codsi lagu jawaabay 401, ama 400 model aan la aqoon Lama tiriyo.
Codsi flood protection diiday Waxaa lagu tiriyaa daaqadda. Waxba lagu ma dallaco.

Codsiyo isku mar ah

Xad ma jiro oo ku saabsan imisa codsi ayuu akoon isku mar furan yeelan karaa, khalad ma jiro codsiyada isku mar loo diro. Codsiyada isla markiiba bilaabmi kari waayay safka ayay ku sugaan oo isku xigxig ayaa laga jawaabaa.

  • Codsi kastaa wuxuu ka tirsan yahay 120 daqiiqad kasta marka uu yimaado, ha dhammaadeen ama yaanay dhammaan codsiyadii hore.
  • Codsi kastaa wuxuu haystaa keydkiisa ilaa uu dhammaado. Labaatan codsi oo furan oo leh miisaaniyadda output-ka default-ka ah waxay haystaan 20 × 4,096 = 81,920 token oo haraag ah. Haddii keydadku wada-jira ka weyn yihiin haraaggaaga, codsiga xiga wuxuu helaa jawaabta Quota exceeded, xataa haddii wicitaannada dhammaaday ay ka jaban lahaayeen. max_tokens yar wuxuu haystaa wax yar.
  • Codsi aan streaming ahayn waxba ma diro ilaa jawaabtiisu dhammaato, sidaas darteed macmiilkaaga sii timeout daboola sugidda. Stream wuxuu isku xirkiisa furan hayaa inta uu sugayo. Streaming

Xadka hal codsi

Xad Qiime Khuseeya Xadka marka la gaaro
Jidhka codsiga 32 MiB (33,554,432 byte) Endpoint kasta Xaalad 413, nooca invalid_request_error.
Miisaaniyadda output-ka: max_tokens, max_completion_tokens, max_output_tokens 1 ilaa 65,536. Default 4,096; shannon-coder-1 /v1/chat/completions iyo /v1/messages default-ku waa 65,536. Model kasta, sida cadadka haraaggaaga laga keydiyo. Sida xadka dhererka jawaabta: model-yada open-weight ee la martigeliyo, shannon-1.6-lite, shannon-1.6-pro iyo shannon-coder-1. Qiime ka baxsan kala duwanaanshaha waxaa loo raraa dhammaadka ugu dhow ee kala duwanaanshaha. Khalad ma jiro.
Stop sequences: stop, stop_sequences 4 string Model-yada open-weight ee la martigeliyo 4-ta string ee madhan ee ugu horreeya ayaa la isticmaalaa.
Sawir ama fayl sida URL loo bixiyay 8 MiB, la akhriyo 20 ilbiriqsi gudahood, ugu badnaan 5 wareejin, cinwaan http ama https oo dadweyne ah Endpoint kasta oo qaata sawirro ama faylal Codsiga waxaa lagu jawaabaa qaybtaas la'aanteed. Khalad ma jiro.
Sawir ama fayl gudaha loo diray (base64) Xad u gaar ah ma laha. Waxay ka tirsan tahay jidhka codsiga 32 MiB. Endpoint kasta oo qaata sawirro ama faylal Xaalad 413 codsiga oo dhan.
text ee POST /v1/tokenize 4,000,000 byte /v1/tokenize Xaalad 413, nooca invalid_request_error, fariinta text too long.
messages ee POST /v1/tokenize iyo jidhka POST /v1/messages/count_tokens Jidhka codsiga 32 MiB Labada endpoint ee tirinta Xaalad 413.
Context window Model kasta: context_window GET /v1/models Model kasta Waxa wada-hadal dheer ku dhaca wuxuu ku xiran yahay model-ka. Model-yo & qiimo
Raadinta webka (web_search: true) Qorshe iyo maalin kasta: Free 3, Plus 30, Standard 50, Pro 60. Hal raadin ayaa la tiriyaa codsi raadintiisu natiijo heshay. Codsiyada dejiya web_search: true Midna la'aan codsiga waxaa lagu jawaabaa raadin la'aan. Khalad ma jiro. Ku-dhismay Raadinta Shabakadda

Khaladaad

Jawaabaha bogga. /v1/messages isla shayga error waxaa loo duubaa sida {"type": "error", "error": {…}}.

Xaalad Nooc Fariin Goorta, iyo waxa la sameeyo
429 rate_limit_error Quota exceeded. Upgrade your plan at shannon-ai.com/plan Keydka codsigu ku soo galin haraaggaaga. Sug 00:00 UTC, kordhi credit, beddel qorshaha, ama dir max_tokens yar.
429 rate_limit_error Too many requests. Retry in <N>s. In ka badan 120 codsi daqiiqadda hadda. Sug N ilbiriqsi oo mar kale dir.
429 rate_limit_error Shannon Coder call quota reached. Upgrade your plan at shannon-ai.com/plan Wicitaannada Shannon Coder ee daaqadda 4-saac ee hadda waa la isticmaalay.
429 rate_limit_error Shannon routes are temporarily busy. Please retry. Model-ku hadda codsiga qaadan karo ma jiro. Mar kale dir kaddib nasasho gaaban.
503 api_error Could not verify your quota right now. Please retry. Haraaggaaga lama akhrin karin. Waxba lagu ma dallaco; mar kale dir codsiga. /v1/responses oo leh model Shannon xaaladdu waa 500.
413 invalid_request_error Jidhka codsigu wuu ka weyn yahay 32 MiB. Endpoint-yada qaabka OpenAI shayga error wuxuu sidaa code: "request_too_large".
413 invalid_request_error text too long text ee POST /v1/tokenize wuu ka dheer yahay 4,000,000 byte.