Total Permintaan (24j)
1,248,592
Kompresi RTK+Caveman
64.2%
Biaya Dihemat
$1,842.50
Rasio Keberhasilan (SLA)
99.991%
Playground AI & Simulator Routing
Ketik prompt untuk mensimulasikan bagaimana OmniRouter memampatkan, mengevaluasi, dan merutekan permintaan secara cerdas.
0 karakter
Klien
Request
OmniRouter
Idle
OpenAI GPT-4o
Primary (120ms)
Claude 3.5 Sonnet
Backup (310ms)
Konsol Terminal OmniRouter
[Sistem] Menunggu input prompt untuk memulai simulasi routing...
Status Provider & Latensi
6 dari 160+ provider utama yang aktif saat ini.
| Model & Provider | Status | Latensi | Biaya (1K tok) |
|---|---|---|---|
|
●
GPT-4o
OpenAI
|
Aktif | $0.0050 | |
|
●
Claude 3.5 Sonnet
Anthropic
|
Aktif | $0.0150 | |
|
●
Gemini 1.5 Pro
Google Cloud
|
Aktif | $0.0070 | |
|
●
Llama 3.1 70B
Groq (LPU)
|
Aktif | $0.0006 | |
|
●
DeepSeek V3
DeepSeek API
|
Saturasi | $0.0002 | |
|
●
Command R+
Cohere
|
Aktif | $0.0030 |
Konfigurasi Aturan Router & Kompresi
Aktifkan modul optimasi pintar untuk mengontrol bandwidth, failover, dan performa query.
Kompresi Token RTK + Caveman
Kompres prompt berulang dan hapus stopwords untuk menghemat hingga 65% token input sebelum dikirim ke API.
Fallback & Auto-Failover Instan
Jika provider utama mengembalikan error 429/5xx, alihkan request secara otomatis ke cadangan terbaik dalam < 50ms.
Routing Cerdas Berbasis Latensi (Dynamic Latency Match)
Secara otomatis memantau performa tiap provider dan memilih rute tercepat berdasarkan ping terbaru.
Model Discovery Custom (Prefiks "kr/")
Sinkronisasi otomatis dengan katalog model Hugging Face dan server model internal kustom.
Informasi Server
Mode Operasi
Unified Proxy API
MCP Server Status
🟢 Connected (4 Tools)
PWA / Cache Status
Active (Offline Mode Ready)
Discovery Catalog Sync
Setiap 1 jam
Koneksi Database
SQLite (Vektor Aktif)
Logs Aktivitas Gateway Real-Time
Pemantauan aktivitas transfer data, failover, dan caching yang sedang berjalan.
[15:00:12] [SYS] OmniRouter Gateway v3.8.11 initialized successfully on port 20128.
[15:00:13] [DB] Connected to database: sqlite-vec active (3,248 cache items loaded).
[15:00:15] [MCP] Registered 4 tools from internal MCP server.
[15:01:40] [ROUTE] POST /v1/chat/completions - strategy: fallback (GPT-4o -> Claude 3.5)
[15:01:40] [RTK] Saved 1,402 tokens via Caveman vocabulary compression (61.5% saved).
[15:01:41] [WARN] GPT-4o returned 429 Too Many Requests (Rate limit hit).
[15:01:41] [FAILOVER] Automatically switching to backup: Claude 3.5 Sonnet.
[15:01:42] [ROUTE] Claude response received successfully in 320ms. Status code 200.
[15:03:01] [ROUTE] POST /v1/chat/completions - strategy: latency (Fastest Llama 3.1 70B selected)
[15:03:01] [ROUTE] Llama response received in 45ms. Status code 200.
[15:04:10] [DISCOVERY] Synced model catalog with Hugging Face registry. Found 4 new "kr/" prefixes.