✦ OpenAI Compatible | Smart Routing Engine

Unified API Gateway, All Models
Smart Routing Cuts Costs 40-80%

ThreeRouter provides a unified AI model access interface, fully compatible with OpenAI SDK. Our L1-L5 smart routing engine automatically matches the optimal cost-performance model based on task complexity, significantly reducing inference costs without compromising quality.

Start Free Trial → Documentation

Core Capabilities

Unified access, smart routing, cost optimization, with compliance as an optional value-added service.

🔌

Unified API Access

Fully compatible with OpenAI API format. Simply change base_url and API Key to switch. No code refactoring required, supporting smooth migration of large-scale microservice architectures.

⚡

L1-L5 Smart Routing

Automatically matches the optimal model tier based on task complexity. Simple tasks use lightweight models, complex reasoning uses top-tier models, achieving optimal cost without sacrificing output quality.

🛡️

Compliance (Optional Add-on)

For enterprises with cross-border needs, we offer EU AI Act compliance reports, audit logs, cross-jurisdictional compliance mapping as optional subscription services.

Call Strategy: One-Click Toggle in Console

Adjust routing strategy anytime from the console. No code changes required. Switch to smart routing to save costs, or specify a fixed model for deterministic output.

🎛️ Console Strategy Settings

🤖 Smart Routing Mode

Automatically selects best cost-performance model

40-80% Cost Reduction

System automatically assigns L1-L5 tiers based on request complexity. Simple requests use lightweight models, complex tasks use premium models.

🎯 Fixed Model Mode

Always uses your preselected model

Deterministic Priority

All requests use a specific model (e.g., GPT-5, GLM-5.1). Ideal for scenarios requiring consistent output characteristics.

# ThreeRouter Console - Routing Strategy Configuration
Current Strategy : Smart Routing (Cost Optimized) Est. Monthly Cost : $2,800 Savings vs Fixed Premium : 52%

# One-click strategy switch (no code changes required)
[ ] Fixed Model:
[✅] Smart Routing (Recommended) - Auto-optimizes cost, stable quality

💡 Tip : Strategy changes take effect immediately. No service restart or API code modification required.

💡 Typical Use Cases: During development and testing, use "Fixed Model Mode" to ensure consistent results. After launch, switch to "Smart Routing Mode" to significantly reduce costs. For A/B testing or compliance audits requiring fixed model outputs, simply toggle back to fixed mode. All changes take effect instantly from the console with zero code changes.

# API call code remains identical regardless of console strategy
client = OpenAI(base_url="https://api.threerouter.ai/v1", api_key="tr-xxx")

# Smart Routing Mode: Auto-optimizes costs, average 40-80% savings
response = client.chat.completions.create(
    model="any",  # Model selection determined by console strategy
    messages=[{"role": "user", "content": "Translate product description"}]
)

# Fixed Model Mode: Always uses your preset model (e.g., GPT-5)
# Same code, zero modifications - strategy toggles instantly in console

L1-L5 Smart Routing Tiers

Automatically matches models based on task complexity, optimizing cost while preserving output quality.

Tier	Model Size	Use Cases	Representative Models	Cost (Input per M tokens)
L1	7B-14B	Text classification, summary extraction, keyword recognition	Qwen3-8B	$0.01
L2	14B-32B	Content generation, translation, customer service replies	Qwen3-32B	$0.05
L3	32B-70B	Complex reasoning, code generation	GLM-5.1	$0.15
L4	70B-235B	Advanced analysis, long-document understanding	Kimi K2.6	$0.30
L5	235B+	Most complex tasks, deep reasoning	DeepSeek V4-Pro / GPT-5.5	$1.10

📊 Validation: A cross-border e-commerce platform processes 80 million monthly API calls, of which 72% are customer service translation and product summary requests. ThreeRouter smart routing automatically assigns these requests to L1/L2 tiers, reducing monthly costs from $42,000 to $16,800 — an 80% reduction — while keeping output quality stable.

Market Background: Model Performance Converges, Price Divergence Intensifies

Top open-source models now approach closed-source model performance, while price gaps reach up to 36x.

Model	Capability Positioning	Price ($/M input)	Relative to Highest Price
GPT-5.5	OpenAI latest flagship	$5.00	Baseline
Claude Opus 4.8	Anthropic latest flagship, SWE-bench Pro 69.2%	$5.00	Baseline
GLM-5.1	Zhipu flagship, SWE-bench first tier	$1.40	72% lower
Kimi K2.6	Open-source SOTA, SWE-Bench Pro 58.6	$0.95	81% lower
DeepSeek V4-Flash	Best value, 284B MoE	$0.14	97% lower
DeepSeek V4-Pro	Flagship reasoning, Codeforces #1	$0.42	92% lower

Massive Cost Optimization Potential

Most enterprise AI calls still use premium closed-source models, yet over 70% of requests don't require top-tier capabilities. Smart routing allocates tasks to appropriate model tiers, achieving significant cost savings while maintaining output quality.

Agent Applications Drive Token Surge

As AI Agent applications become ubiquitous, single-task token consumption has increased 50-200x compared to traditional Bot patterns. In coding scenarios, power users consume 200M+ tokens monthly — model selection dramatically impacts annual costs.

40-80%

Avg Smart Routing Cost Reduction

5-18x

Model Price Gap Multiplier

30%+

Standard Plan vs Official Pricing

16.9T

Weekly Industry Token Volume (Mar 2026)

Pricing Plans

Standard plan already 30%+ below official pricing, with compliance as an optional add-on.

Plan	Models	Price (Input per M tokens)	Description
Standard	DeepSeek V4 / GLM-5.1 / Kimi K2.6	$0.22 – 0.87	30%+ below official pricing. For general scenarios without cross-border compliance needs.
Compliance	DeepSeek V4 / GLM-5.1 / Kimi K2.6	$0.38 – 1.50	Includes audit logs, EU AI Act compliance tags. ~20% premium over official pricing.
Premium	GPT-5 / Claude Opus 4.6	$5.00 – 5.80	Supports smart routing and compliance auditing.

Value-Added Service (Compliance)	Monthly Fee	Features
Starter Compliance	¥3,500 / mo	Audit logs, model risk labels, basic compliance reports
Business Compliance	¥14,000 / mo	EU AI Act automated compliance reports, cross-jurisdictional compliance mapping
Enterprise Custom	¥35,000+ / mo	Private deployment, dedicated compliance consulting, SLA guarantees

🎯 New User Benefit: Register for 1M free tokens to experience smart routing cost optimization. Compliance add-ons are optional — only pay if you need them.

Cost Advantage Sources

📉 Compute Subsidy Policy

ThreeRouter's Beijing tech center benefits from Haidian District compute subsidies (30% of contract value, up to ¥20M/year) and municipal compute vouchers (20% of contract value, up to ¥2M/year). These savings are passed directly to customers.

🧠 Routing Optimization

L1-L5 dynamic tiering ensures simple tasks use low-cost models, avoiding over-reliance on premium models. Comprehensive analysis shows routing optimization delivers an additional 20-30% inference cost reduction.

Competitive Comparison

Dimension	OpenRouter	Direct API Calls	ThreeRouter
Pricing Strategy	Official or slightly higher	Official pricing	Standard plan 30%+ below official
Cost Optimization	No smart routing	No	L1-L5 smart routing, 40-80% savings
Strategy Flexibility	Model specified in code	Per-request model selection	One-click console toggle, zero code changes
Compliance	None	None	Optional add-on (subscribe as needed)

Typical Use Cases

🌏 Cross-Border E-commerce

Multi-language product descriptions, intelligent customer service translation, marketing content localization. Over 80% of these requests are handled with high quality by L1/L2 models, achieving 50%+ cost reduction.

🤖 Code Assistant / AI Agent

Code generation, auto-completion, Agent task orchestration. Power users consume 200M+ tokens monthly — model selection directly impacts annual costs.

📦 Global SaaS

SaaS products targeting overseas markets can leverage the Standard plan for cost savings; subscribe to Compliance add-ons when needed for EU AI Act or other regulatory requirements.

Quick Start

# Step 1: Register and get API Key
export THREE_ROUTER_API_KEY="tr-xxx"

# Step 2: Call via unified gateway (OpenAI SDK compatible)
curl https://api.threerouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THREE_ROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "any",
    "messages": [{"role": "user", "content": "Generate coffee product marketing copy for Southeast Asian markets"}]
  }'

# Step 3: One-click toggle routing strategy in console (Smart / Fixed)
# No code changes required - strategy changes take effect immediately

# Optional: Enable compliance (requires subscription)
# -H "X-Compliance-Region: EU"

Free Trial → Documentation

Unified API Gateway, All ModelsSmart Routing Cuts Costs 40-80%