Nyaraka za API
Miundombinu ya Akili Isiyodhibitiwa
API ya AI isiyodhibitiwa: model za Shannon zisizokataa na zisizo na vichujio, zinazotolewa kutoka GPU zetu wenyewe. Zinaoana na OpenAI na Anthropic — tumia SDK unayotumia tayari, key moja, streaming, tool calling na reasoning kwenye kila endpoint.
- 22 model zisizodhibitiwa
- 256K dirisha la muktadha
- kuanzia $0.50/1M
- 3 API zinazooana
- 3 mitandao: TLS, Tor, I2P
Anza kujenga Jaribu Playground
from openai import OpenAI
client = OpenAI(
api_key="YOUR_API_KEY",
base_url="https://api.shannon-ai.com/v1"
)
response = client.chat.completions.create(
model="shannon-3",
messages=[
{"role": "user", "content": "Hello, Shannon!"}
],
max_tokens=1024
)
print(response.choices[0].message.content) import OpenAI from 'openai';
const client = new OpenAI({
apiKey: 'YOUR_API_KEY',
baseURL: 'https://api.shannon-ai.com/v1'
});
const response = await client.chat.completions.create({
model: 'shannon-3',
messages: [
{ role: 'user', content: 'Hello, Shannon!' }
],
max_tokens: 1024
});
console.log(response.choices[0].message.content); package main
import (
"context"
"fmt"
openai "github.com/sashabaranov/go-openai"
)
func main() {
config := openai.DefaultConfig("YOUR_API_KEY")
config.BaseURL = "https://api.shannon-ai.com/v1"
client := openai.NewClientWithConfig(config)
resp, err := client.CreateChatCompletion(
context.Background(),
openai.ChatCompletionRequest{
Model: "shannon-3",
Messages: []openai.ChatCompletionMessage{
{Role: "user", Content: "Hello, Shannon!"},
},
MaxTokens: 1024,
},
)
if err != nil {
panic(err)
}
fmt.Println(resp.Choices[0].Message.Content)
} curl -X POST "https://api.shannon-ai.com/v1/chat/completions" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "shannon-3",
"messages": [
{"role": "user", "content": "Hello, Shannon!"}
],
"max_tokens": 1024
}' Model na bei KWA TOKENS 1M
Mipaka, id na jinsi bei zinavyopangwaModel za open-weight zinazopangishwa muktadha wa tokens 262,144 kwenye kila id; input na output zinatozwa kando; kila id inakubali kile kinachokubaliwa na model inayoiga
- Isiyodhibitiwa
DeepSeek-V4-Pro-0813
3BIT-REAPDeepSeek · 1.6T MoE · 49B active
Model ya daraja la juu kwa reasoning ya kina na output ndefu.
Model idDeepSeek-V4-Pro-0813-3BIT-REAP- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $1.95 / $0.488 / $3.90
- Isiyodhibitiwa
GLM-5.2
3BIT-REAPZ.ai · 744B MoE · 40B active
Model imara ya matumizi mengi yenye output tajiri ya lugha nyingi.
Model idGLM-5.2-3BIT-REAP- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.73 / $0.183 / $2.34
- Isiyodhibitiwa
Kimi-K3
3BIT-REAPMoonshot AI · 2.8T MoE · 104B active
Model ya hali ya juu iliyorekebishwa kwa majibu ya kina na ya kina.
Model idKimi-K3-3BIT-REAP- Muktadha
- 256K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $3.83 / $0.958 / $19.12
- Isiyodhibitiwa
Nemotron3Ultra
3BIT-REAPNVIDIA · 550B hybrid Mamba-Attention MoE · 55B active
Model kubwa yenye ufanisi na reasoning imara kwa kila dola.
Model idNemotron3Ultra-3BIT-REAP- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.75 / $0.188 / $3.30
- Isiyodhibitiwa
MiniMax-M3
3BIT-REAPMiniMax · 428B MoE · 23B active · sparse attention
Model ya haraka na ya gharama ya chini kwa mzigo mkubwa wa kazi.
Model idMiniMax-M3-3BIT-REAP- Muktadha
- 256K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.50 / $0.125 / $2.00
- Isiyodhibitiwa
DeepSeek-V4-Flash-0731
W4A16-AUTOROUND-REAPDeepSeek · 284B MoE · 13B active
Model ya haraka ya DeepSeek inayosawazisha kasi na ubora.
Model idDeepSeek-V4-Flash-0731-W4A16-AUTOROUND-REAP- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.50 / $0.125 / $2.00
- Isiyodhibitiwa
Kimi-K2.6
W4A16-AUTOROUND-REAPMoonshot AI · 1T MoE · 32B active
Model yenye uwezo kwa reasoning na uandishi wa rasimu wa kila siku.
Model idKimi-K2.6-W4A16-AUTOROUND-REAP- Muktadha
- 256K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.78 / $0.195 / $3.67
- Isiyodhibitiwa
Laguna-S-2.1
W4A16-AUTOROUND-REAPPoolside · 118B MoE · 8B active
Model ya gharama ya chini kabisa kwa kazi nyepesi.
Model idLaguna-S-2.1-W4A16-AUTOROUND-REAP- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- Hakuna matokeo yaliyopangwa
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.50 / $0.125 / $2.00
- Isiyodhibitiwa
inkling
W4A16-AUTOROUND-REAPThinking Machines · 975B MoE · 41B active
Model ya ubunifu yenye mtindo wa kipekee wa uandishi.
Model idinkling-W4A16-AUTOROUND-REAP- Muktadha
- 256K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- Kitu cha JSON
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $1.42 / $0.355 / $6.07
- Isiyodhibitiwa
MiMo-V2.5-Pro
W8A16Xiaomi · 1.02T MoE · 42B active · 8-bit
Model ndogo ya pro kwa output iliyopangwa na sahihi.
Model idMiMo-V2.5-Pro-W8A16- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.50 / $0.125 / $2.00
- Isiyodhibitiwa
MiMo-V2.5
W8A16Xiaomi · 8-bit multimodal · W8A16
Model ndogo na ya haraka kwa kazi rahisi kwa gharama ya chini kabisa.
Model idMiMo-V2.5-W8A16- Muktadha
- 256K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.50 / $0.125 / $2.00
- Isiyodhibitiwa
Hy3
W8A16Tencent · 295B MoE · 21B active · 8-bit
Model ya bei nafuu kwa majibu mafupi ya mazungumzo.
Model idHy3-W8A16- Muktadha
- 256K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Reasoning effort
- Prompt caching
- Streaming
- Ndani / cached / nje kwa 1M
- $0.50 / $0.125 / $2.00
Model za Shannon bei moja isiyobadilika kwa input na output
-
IsiyodhibitiwaShannon 1.6 Lite
Majibu ya haraka na yenye ufanisi kwa kazi za kila siku
Model idshannon-1.6-lite- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Web search
- Streaming
- Bei / tokens 1M
- $3.90
-
IsiyodhibitiwaShannon 1.6 Pro
Reasoning ya hali ya juu kwa matatizo magumu
Model idshannon-1.6-pro- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $7.80
-
IsiyodhibitiwaShannon 2 Lite
Shannon 2 Lite: pipeline ya haraka ya kufikiri-kuandaa-kukagua yenye ufuatiliaji wa reasoning unaoonekana. Chat ya kila siku, muhtasari na Maswali na Majibu.
Model idshannon-2-lite- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $3.90
-
IsiyodhibitiwaShannon 2 Pro
Shannon 2 Pro: hatua za kina zaidi za kufikiri-kuandaa-kukagua kwa uchambuzi, uandishi mrefu na reasoning ya hatua nyingi.
Model idshannon-2-pro- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $5.85
-
IsiyodhibitiwaShannon 3
Model ya Shannon ya kizazi cha hivi karibuni yenye reasoning ya haraka kwa bei ya chini isiyobadilika.
Model idshannon-3- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $3.35
-
IsiyodhibitiwaShannon 3 Pro
Kiwango cha Shannon 3 chenye uwezo wa juu zaidi kwa kazi ngumu za hatua nyingi.
Model idshannon-3-pro- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $3.35
-
IsiyodhibitiwaShannon 3.1
Shannon 3 kwenye vifaa vyetu wenyewe, ikitiririka kwa kasi kamili bila kizuizi cha kasi.
Model idshannon-3.1- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $3.35
-
IsiyodhibitiwaShannon 3.1 Pro
Reasoning ya kina kwa kasi kamili pamoja na hatua ya utafiti wa nyuma kwa maswali magumu.
Model idshannon-3.1-pro- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Web search
- Streaming
- Bei / tokens 1M
- $3.35
-
IsiyodhibitiwaShannon 3.5
One streaming answer that thinks, searches, draws charts and diagrams and writes files inline.
Model idshannon-3.5- Muktadha
- 192K
- Input
- Maandishi + picha kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Streaming
- Bei / tokens 1M
- $3.35
-
IsiyodhibitiwaShannon Coder 1
Mtaalamu wa code na tools: uhariri wa agentic, mafaili marefu na mizunguko mikali ya tools.
Model idshannon-coder-1- Muktadha
- 128K
- Input
- Maandishi kama input
- Matokeo yaliyopangwa
- JSON schema
- Tools
- Reasoning
- Streaming
- Bei / tokens 1M
- $8.00
Pitia nyaraka
- Mwanzo wa Haraka Unda key, elekeza SDK yako kwa Shannon na utume ombi la kwanza.
- Model na bei Kila model id pamoja na dirisha lake la muktadha, input na bei kwa tokens 1M.
- Uthibitishaji Key moja kwa kila endpoint: header ya Bearer au x-api-key.
- Kutiririsha Server-sent events zenye reasoning deltas; matumizi hufika kwenye chunk ya mwisho.
- Uitoaji wa kazi Fafanua tools, acha model ziite, ziongoze kwa tool_choice.
- Matokeo yaliyopangwa Majibu kama kitu cha JSON au kulingana na JSON schema yako.
- Reasoning effort Soma mtiririko wa reasoning; weka effort kwenye model za open-weight zinazopangishwa.
- Kukesha kwa prompt Model zinazopangishwa hukesha kiotomatiki mwanzo wa prompt unaorudiwa, kwa 25% ya bei ya input.
- Responses API Lahaja ya OpenAI Responses kwenye model zile zile na key ile ile.
- Anthropic Endelea na SDK yako ya Anthropic na code ya Messages; badilisha base URL tu.
- Zana za CLI za kuandika code Claude Code na Codex CLI zikiwa na Shannon kama model nyuma yake.
- Ushughulikiaji wa Makosa Hali za HTTP, aina za makosa na umbo la mwili wa kosa.
Uko tayari kujenga?
Kiwango cha bure kimejumuishwa kwenye kila akaunti. Key yako iko umbali wa kubofya mara moja.