Atria-Dawn-Preview (256K Context) Port: 7842
🔒 Context Preservation Mode: Active Key is Key #4
All incoming requests stick to Key #4 to keep KV prompt cache hot and preserve context. If this key is rate-limited, it enters a 35-minute cooldown and advances to the next key. If it is exhausted (100M tokens), it is permanently retired.
Active Keys
7 / 7
Exhausted Keys
0
Total Tokens Used
201,580
Remaining Allowance
699,798,420

Keys Priority Queue

(Cooldown: 35 mins, Max Retries: 5)
Key Name Masked Key Live Status & Speed Token Usage (Actual / 100M) Remaining Tokens Actions
Key #1 atr_...l4wH ACTIVE⚡ 2417.5msRPM: 49/50
15,527 tokens 0.0155%
99,984,473
Key #2 atr_...Xuft ACTIVE⚡ 4021.9msRPM: 49/50
63,610 tokens 0.0636%
99,936,390
Key #3 atr_...yKHG ACTIVE⚡ 15704.7msRPM: 48/50
48,622 tokens 0.0486%
99,951,378
⭐ PRIMARY KEYKey #4 atr_...CCNb ACTIVE⚡ 5750.1msRPM: 49/50
15,614 tokens 0.0156%
99,984,386
Key #5 atr_...29va ACTIVE⚡ 5197.9msRPM: 49/50
18,287 tokens 0.0183%
99,981,713
Key #6 atr_...57_w ACTIVE⚡ 5197.8msRPM: 49/50
19,230 tokens 0.0192%
99,980,770
Key #7 atr_...FmxQ ACTIVE⚡ 20515.1msRPM: 49/50
20,690 tokens 0.0207%
99,979,310

🚀 Client Connection Guide (No Auth Required)

Any client can connect with any random string as API key. The proxy transparently routes through the active primary key.

1. OpenAI SDK / Custom Tools (Cursor, Open WebUI, NextChat)

Base URL: http://localhost:7842/v1
API Key:  sk-any-random-string
Model:    Atria-Dawn-Preview

2. Claude Code (Anthropic Messages API)

export ANTHROPIC_BASE_URL="http://localhost:7842"
export ANTHROPIC_AUTH_TOKEN="any-random-string"
claude --model Atria-Dawn-Preview