Docsone key · every model · OpenAI-compatible

Integrate every model in minutes

Keep the OpenAI SDK you already use. Point it at NexusAI, swap the key, and call GPT, Claude, Gemini, DeepSeek, Qwen, GLM, Kimi or Doubao by changing one string.

One key, every model

Drop-in OpenAI-compatible API. Keep your SDK, change the base URL and key, then call any model — Western or Chinese — by its id.

Quickstart

Authenticate once.
Reach every provider.

No per-vendor accounts, no Chinese phone verification, no separate billing. One NexusAI key authenticates GPT, Claude, Gemini, Grok and DeepSeek, Qwen, GLM, Kimi, Doubao behind a single global endpoint.

Base URL

https://api.nexusai.com/v1

Auth header

Authorization: Bearer sk-nexus-•••••••••

SDK

openai · @anthropic-ai/sdk · any OpenAI client

Model IDs

gpt-5.1OpenAI · Flagship multimodal

claude-opus-4.5Anthropic · Long-context agentic

gemini-3-proGoogle · 1M-token context

grok-4xAI · Real-time knowledge

deepseek-v3.2DeepSeek · 深度求索 · Cost-efficient MoE

qwen3-maxAlibaba · 通义千问 · Multilingual, 100+ langs

glm-4.6Zhipu · 智谱 · Strong Chinese reasoning

kimi-k2Moonshot · 月之暗面 · Very long context

doubao-proByteDance · 字节豆包 · Low-latency, high-volume

deepseek-r1DeepSeek · 推理 · Open reasoning

curl https://api.nexusai.com/v1/chat/completions \
  -H "Authorization: Bearer $NEXUSAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v3.2",
    "messages": [
      { "role": "user", "content": "Explain MoE in one sentence." }
    ]
  }'

# Switch provider by changing ONE string — same key, same endpoint:
#   "model": "gpt-5.1"          -> OpenAI
#   "model": "claude-opus-4.5"  -> Anthropic
#   "model": "kimi-k2"          -> Moonshot 月之暗面
#   "model": "doubao-pro"       -> ByteDance 字节豆包
#   "model": "qwen3-max"        -> Alibaba 通义千问

Command every model from one place

Real-time token consumption, spend and latency per model — Western and Chinese — on a single dashboard. One key, total visibility, no surprises.

$866.80

Spend this month

−18% vs last month

4.92M

Requests (30d)

across 6 models

412 ms

Avg. latency

p50, routed

1,043

Failovers handled

0 dropped requests

ModelTokensSpendShare of spend

deepseek-v3.2

DeepSeek · 深度求索

184.2M$312.4

34%

gpt-5.1

OpenAI

92.7M$268.9

29%

claude-opus-4.5

Anthropic

61.0M$173.5

19%

qwen3-max

Alibaba · 通义千问

74.8M$58.2

kimi-k2

Moonshot · 月之暗面

40.3M$31.7

doubao-pro

ByteDance · 字节豆包

120.5M$22.1

Set per-model budgets and spend alerts. Throttle, route around, or hard-cap any provider — without touching your code.

Every model, one endpoint

Switch between GPT, Claude, Gemini, Grok and the leading Chinese models — DeepSeek, Qwen, GLM, Kimi, Doubao — by changing a single string. Same OpenAI-compatible API.

GPT-5.1

Flagship

OpenAI's flagship multimodal model. Best-in-class general reasoning, coding, and tool use.

Integrate every model in minutes

One key, every model

Authenticate once.Reach every provider.

Command every model from one place

Every model, one endpoint

GPT-5.1

Claude Opus 4.5

Gemini 3 Pro

Grok 4

OpenAI o4

Claude Opus 4.5 (Thinking)

DeepSeek R1

Gemini 3 Deep Think

DeepSeek V3.2

Qwen3 Max

GLM-4.6

Kimi K2

Doubao Pro

Hunyuan

ERNIE 4.5

MiniMax M2

Llama 4

Mistral Large

Qwen3 (Open Weights)

DeepSeek V3 (Open Weights)

Ship with any model in minutes

Authenticate once.
Reach every provider.