# Kai — Decision Intelligence — Use via Hanzo AI API > Hanzo Kai is a structured decision model answering typed questions at /v1/d... Call via Hanzo's OpenAI-compatible API. Context: 33K. [Models](https://hanzo.ai/models)/[Hanzo](https://hanzo.ai/models/hanzo)/kai-1.4 # Kai [Hanzo](https://hanzo.ai/models/hanzo)Enterprise Hanzo Kai is a structured decision model answering typed questions at /v1/decisions with calibrated probabilities and zero output tokens. Text core shipping (322M params, 32K reach); multimodal vision and audio towers in preview. TextStateVision (preview)Audio (preview) [Get API Key](https://platform.hanzo.ai/api-keys)[View Docs](https://docs.hanzo.ai)[Explore Kai →](https://hanzo.ai/kai)[GitHub ↗](https://github.com/hanzoai/kai) ## Specifications Model ID kai-1.4 Live Checkpoint kai-1.3 (pin @551f1990…) Architecture 22-layer ModernBERT encoder + factorized cross-attention Parameters 322.1M (text decision core, F32) Context Reach 32,768 tokens (YaRN; trained: 1,024) Qualified Modality Text (Vision, Audio, Video in preview) Endpoint POST /v1/decisions Output Tokens 0 (typed probability distribution) Latency (CPU) 85–95 ms p50 Safety Qualification Zoo Lifeguard framework Price … Status available ### Features - Live: High-throughput text & structured state decision core - Live: Calibrated probability distributions over typed options (choice, noul, score) - Live: 85–95 ms p50 CPU latency (4 cores, sequence length up to 1,024 context) - Live: 32,768 positional reach via YaRN rotary stretching (1,024 trained base) - Live: Zero output tokens generated or billed ($0.021/MTok base input) - Live: Integrated Zoo Lifeguard safety & harm-prevention alignment - Preview: Multimodal vision, audio, and video decision towers (~850M total arch) - Preview: Dynamic adaptive reasoning depth (reasoning: "auto" | "fast") ## Quick Start Instant Guest API Keyguest_temp_key No account or sign up required Test right now from your terminal. No credentials needed: cURL · instant guest request Copy ``` curl https://api.hanzo.ai/v1/chat/completions \ -H "Authorization: Bearer guest_temp_key" \ -d '{"model": "zen-free", "messages": [{"role": "user", "content": "hello"}]}' ``` TypeScript ``` import { HanzoAI } from '@hanzo/ai' const ai = new HanzoAI({ apiKey: process.env.HANZO_API_KEY, baseURL: 'https://api.hanzo.ai/v1' }) // Decisions, not completions: typed evaluation over state with zero output tokens const decision = await ai.postDecisions({ model: 'kai', state: 'System telemetry: Cluster node-04 memory pressure 92%, replica pool degraded.', questions: [ { question: 'Should we failover traffic to availability zone B?', kind: 'boolean', instructions: 'Return true to execute failover, false to hold.' } ] }) console.log(decision.answers[0]) ``` Python ``` import os import requests # Decisions, not completions: POST /v1/decisions with typed questions response = requests.post( "https://api.hanzo.ai/v1/decisions", headers={ "Authorization": f"Bearer {os.environ['HANZO_API_KEY']}", "Content-Type": "application/json" }, json={ "model": "kai", "state": "System telemetry: Cluster node-04 memory pressure 92%, replica pool degraded.", "questions": [ { "question": "Should we failover traffic to availability zone B?", "kind": "boolean", "instructions": "Return true to execute failover, false to hold." } ] } ) print(response.json()["answers"][0]) ``` cURL ``` curl https://api.hanzo.ai/v1/decisions \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $HANZO_API_KEY" \ -d '{ "model": "kai", "state": "System telemetry: Cluster node-04 memory pressure 92%, replica pool degraded.", "questions": [ { "question": "Should we failover traffic to availability zone B?", "kind": "boolean", "instructions": "Return true to execute failover, false to hold." } ] }' ``` Go ``` package main import ( "bytes" "encoding/json" "fmt" "net/http" "os" ) func main() { payload := map[string]interface{}{ "model": "kai", "state": "System telemetry: Cluster node-04 memory pressure 92%, replica pool degraded.", "questions": []map[string]interface{}{ { "question": "Should we failover traffic to availability zone B?", "kind": "boolean", }, }, } body, _ := json.Marshal(payload) req, _ := http.NewRequest("POST", "https://api.hanzo.ai/v1/decisions", bytes.NewBuffer(body)) req.Header.Set("Authorization", "Bearer "+os.Getenv("HANZO_API_KEY")) req.Header.Set("Content-Type", "application/json") client := &http.Client{} resp, err := client.Do(req) if err != nil { panic(err) } defer resp.Body.Close() var result map[string]interface{} json.NewDecoder(resp.Body).Decode(&result) fmt.Println(result["answers"]) } ``` ## More from Hanzo [Enso900KHanzo's proprietary frontier model — Opus-class reasoning by default with 1M-context overflow.](https://hanzo.ai/models/hanzo/enso)[Enso900KRoutes each request to one tier by the reasoning depth it asks for: enso-flash at off, none, minimal, low, default; enso-pro at medium, high; enso-ultra at xhigh, max. It bills at the price of the tier that answers; the price shown is a request's that states no depth. When nothing pays for that tier it answers on enso-free, at no charge, and says so.](https://hanzo.ai/models/hanzo/enso-auto)[Enso Flash900KFast, economical Enso tier for high-volume, low-latency everyday work, with 1M-context overflow.](https://hanzo.ai/models/hanzo/enso-flash)[enso-free1M](https://hanzo.ai/models/hanzo/enso-free)[enso-pro1M](https://hanzo.ai/models/hanzo/enso-pro)[Enso Ultra1MAdaptive fan-out — probes a task-appropriate model, escalates to a top-K panel only when needed, then verifies-then-selects the best answer.](https://hanzo.ai/models/hanzo/enso-ultra) [View all Hanzo models →](https://hanzo.ai/models/hanzo) ## Use Kai via Hanzo AI One API key. Every major model. OpenAI-compatible. Start free. [Get Free API Key](https://platform.hanzo.ai/api-keys)[Browse All Models](https://hanzo.ai/models)