Skip to content
Intelligent Routing
Smart LLM Routing
Automatically route requests to the best model based on cost, latency, and availability. One API, 600+ models.
01Routing
Routing that adapts in real time
Failover, region pinning, and per-team policies decide where every request lands. No code changes when a provider degrades.
Failover Monitor14:23:47 UTC
| Time | Event |
|---|---|
| 14:23:47 | OpenAI gpt-5.4 endpoint degraded |
| 14:23:47 | Failover initiated to Anthropic opus-4.6 |
| 14:23:47 | Failover complete in 18ms |
| 14:23:48 | All traffic routed to Anthropic |
Provider HealthOpenAI: DegradedAnthropic: HealthyGoogle: Healthy
Request RoutingLIVE
| Region | Serves | Latency, requests |
|---|---|---|
| eu-west-1Primary | Frankfurt | 12ms, 14.2K/min |
| us-east-1Active | Virginia | 8ms, 22.1K/min |
| ap-southeast-1Active | Singapore | 18ms, 6.8K/min |
Data Residency EnforcedEU requests to eu-west-1 only
Policy EngineORG TO USER
| Scope | Budget | Used |
|---|---|---|
| Org Global Spending Limit | $21,000 / $50,000 | 42% |
| Team Engineering | $3,250 / $5,000 | 65% |
Usersarah@acme.com
InheritsRate Limit: 100 req/minRegion: EU
Agent Routing5 AGENTS
support-triage$0.002/req cap
primary deepseek-r3fallback llama-4Cost-optimized
code-reviewer$0.05/req cap
primary opus-4.6fallback gpt-5.4Quality-first
data-extractor$0.01/req cap
primary gpt-5.4fallback gemini-3.1-proBalanced
content-writer$0.04/req cap
primary opus-4.6fallback gemini-3.1-proQuality-first
search-agent$0.005/req cap
primary gemini-3.1-profallback deepseek-r3Latency-first
02Integrate
One Line to Smart Routing
Change one line of code to start using intelligent routing
terminalcURL
# Replace: https://api.openai.com/v1
# With: https://router.requesty.ai/v1
curl https://router.requesty.ai/v1/chat/completions \
-H "Authorization: Bearer $REQUESTY_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "anthropic/claude-opus-4-6", "messages": [...]}'- Auto Failover
- Geo-Routing
- Custom Policies