The self-improving intelligence network for AI agents
Route every task to the best model automatically. AgentOptima tracks real performance across models, learns from results, and continuously improves how your agent thinks.
Sign up for AgentOptima to get your API key. Every request authenticates
with your key in the header.
Header:X-API-Key: ao-xxxxx
Currently in beta — contact us for access
3
Plug into your agent
Add AgentOptima to your agent's workflow. Classify tasks and get model recommendations
with two simple API calls.
import requests
# Classify your task
response = requests.post(
"https://agentoptima.ai/api/v1/classify",
headers={"X-API-Key": "your-key"},
json={"message": "your task description"}
)
result = response.json()
# → {"task_type": "coding", "recommended_cascade": ["gpt-4o-mini", "claude-haiku", "o3-mini"]}# Get model recommendation
rec = requests.get(
"https://agentoptima.ai/api/v1/recommend?task_type=coding",
headers={"X-API-Key": "your-key"}
).json()
# → Use rec["recommended_model"] for your task
4
Watch costs drop
AgentOptima routes each task to the most cost-effective model.
Most teams see 60-80% cost reduction within the first week.
💰 60-80% cost reduction
⚡ Quick API Reference
Four endpoints to integrate AgentOptima into any agent workflow.
POST/api/v1/classify
Classify a task and get model cascade
GET/api/v1/recommend
Get best model for task type
GET/api/v1/cascade
Get full retry waterfall
POST/api/v1/track
Log task outcomes for learning
🏆 Live Model Rankings
🟢 Data-driven routing active — AgentOptima Score combines quality, reliability and cost efficiency across 400K+ real inference outcomes.
Loading rankings…
🗺️ Routing Map
Every task type routes to the best model — backed by 400K+ real task outcomes across live agents, research benchmarks, and routing data. Send any task to /api/v1/recommend and get the optimal model instantly.
Loading routing map…
⚡ Live Task Feed
Most recent tasks logged across all agents and models. Updates every 15s.
Loading recent tasks…
🧠 Routing Intelligence
Live efficiency data from /api/v1/efficiency — best-value models vs Sonnet baseline, updated every 60s.
Best Value Models
Loading efficiency data…
📋 Routing Recommendations
⏳ Loading recommendations…
🧠 Session Intelligence
Real-time context-aware model routing savings. Aris shifts to cheaper models when context is large, tracking every dollar saved.
--
SI Savings
estimated via model shifts
--
Shifts
total model shifts executed
--
Top Model
most-shifted-to model
Loading Session Intelligence data…
📊 Task State Board
Live task health across all agents. Tracks success, failures, retry chains, and error types.
⚠️ Recent Failures
Auto-refreshes every 30s
📋 Worker Contracts
Typed input/output schemas per task type. Workers receive a contract spec — not a vague prompt.
🔌 Connect Your Agent
External agent access is coming soon.
🔒
Coming Soon
We're finalising security audits and access controls before opening external agent registration.
Early access will be invite-only.