14-Dimension Smart Routing

Claude Code, Smarter Routing

14-dimension smart routing across 9 Claude models. Bedrock/Vertex/Anthropic failover. Native Anthropic protocol.

terminal

# Get your API key at chuizi.ai

# Set two environment variables

$ export ANTHROPIC_BASE_URL=https://api.chuizi.ai/anthropic

$ export ANTHROPIC_API_KEY=ck-your-key-here

# Done. Claude Code now uses Chuizi.AI

$ claude "build me a landing page"

Intelligent Routing, Automatic Savings

Most Claude Code requests are simple: file reads, small edits, linting, git operations. Our 14-dimension scorer evaluates every request in under 1ms and routes to the optimal model. Simple tasks go to Haiku ($1/$5 per M tokens), daily coding to Sonnet ($3/$15), complex reasoning to Opus ($15/$75). Requests are routed through Bedrock (primary), Vertex, and Anthropic direct with automatic failover.

Request
Chuizi.AI Gateway

14-Dimension Scorer

<1ms

LIGHT

score < 0.0

Claude Haiku

$1 / $5 per M

MEDIUM

0.0 - 0.35

Claude Sonnet

$3 / $15 per M

HEAVY

score > 0.35

Claude Opus

$15 / $75 per M

14-Dimension Request Scoring

Token Count

Long context auto-upgrades to more capable models

Code Presence

Detects code generation and editing needs

Reasoning Markers

Identifies 'think step by step', 'analyze' patterns

Technical Terms

Measures specialized vocabulary density

Creative Markers

Detects creative writing and brainstorming

Simple Indicators

Yes/no questions, translations, summaries

Multi-step Patterns

Detects multi-step task requirements

Question Complexity

Evaluates overall question difficulty

Imperative Verbs

Measures command verb density and urgency

Constraints

Counts specific requirements and restrictions

Output Format

Detects structured output needs (JSON, tables)

Domain Specificity

Assesses domain expertise requirements

Agentic Tasks

Identifies tool calling and agent workflows

Relay Indicators

Simple relay, forward, or passthrough detection

See Your Savings

10,000
1,000100,000

Always Opus

$1.1k

/month

Always Sonnet

$210

/month

Chuizi.AI Smart

$235

/month

You save $815/month vs always-Opus

(78% less than always-Opus pricing)

Chuizi.AI vs Alternatives

FeatureAnthropic DirectAWS BedrockChuizi.AI
PricingStandardSame as AnthropicUpstream + 5% markup
Smart RoutingNoneNone3-tier (Haiku/Sonnet/Opus)
Auto FailoverNoneCross-region onlyBedrock → Vertex → Anthropic
Setup Time1 min30 min (IAM, region)30 seconds
Multi-protocolAnthropic onlyConverse APIOpenAI + Anthropic + Gemini
Chinese ModelsNoneNoneFull (Qwen, DeepSeek, Doubao)
WeChat Pay
Prompt CachingPassthroughPassthroughPassthrough
Routing SavingsNone (single model)None (single model)Via smart routing (Haiku for simple tasks)

Supported Claude Models

All Claude models routed through Bedrock (primary), Vertex, and Anthropic direct with automatic failover. Prices shown are upstream per-million-token rates; chuizi.ai adds a 5% markup.

Claude Opus 4.6

HEAVY

Latest flagship -- complex reasoning & architecture

Input: $15/MOutput: $75/M

Claude Opus 4.5

HEAVY

Previous-gen flagship reasoning model

Input: $15/MOutput: $75/M

Claude Opus 4.1

HEAVY

Stable Opus for long-running tasks

Input: $15/MOutput: $75/M

Claude Sonnet 4.6

MEDIUM

Latest Sonnet -- daily coding & debugging

Input: $3/MOutput: $15/M

Claude Sonnet 4.5

MEDIUM

Previous-gen balanced performance

Input: $3/MOutput: $15/M

Claude Sonnet 4

MEDIUM

Cost-effective coding model

Input: $3/MOutput: $15/M

Claude Sonnet 4 Thinking

MEDIUM

Extended thinking for complex problems

Input: $3/MOutput: $15/M

Claude Haiku 4.5

LIGHT

Fast & efficient for simple tasks

Input: $1/MOutput: $5/M

Claude Haiku 3.5

LIGHT

Ultra-fast, lowest cost option

Input: $0.25/MOutput: $1.25/M

Frequently Asked Questions

Start Saving in 30 Seconds

$0.50 free credits included. No credit card required.