Welcome to DirectRouter
Inference Controls
Curated Model Catalog 7 EDGE MODELS
Sovereign routing across Anthropic, OpenAI, Google DeepMind, and open-source models with official pass-through rates.
Earn Free Credits +$10.00 REWARDS
Complete the daily sponsor step for $10.00 and invite developers for $10.00 per friend.
Daily Linkvertise Reward
Click the Linkvertise link, complete the brief sponsor step, and claim +$10.00 in credits directly to your balance. Resets every 24 hours.
Invite Friends, Earn $10.00 & Unlock Models
Share your personal invite link. When a new user registers with your link, you instantly receive +$10.00 in credits. Claude 5 Opus is unlocked by default, inviting 1 friend unlocks GPT 6 Astra, and inviting 2 friends unlocks Claude 5.5 Opus!
API Key Management SECRET KEYS
Provision secure bearer keys for OpenAI SDK, Cursor, Claude Code, and terminal agents.
| Name | Key Prefix | Created | Last Used | Actions |
|---|
Endpoint Connection Settings
DirectRouter is a drop-in replacement for OpenAI endpoints:
Developer Docs & Quickstart v2.0
Plug DirectRouter into Claude Code, Codex, Cursor, Python, and terminal agents.
Anthropic Claude Code CLI Native /v1/messages
DirectRouter provides built-in emulation of Anthropic's Messages API (/v1/messages). You can run Anthropic's official Claude Code command-line tool with zero proxies or wrappers.
export ANTHROPIC_BASE_URL="{BASE_URL}"
export ANTHROPIC_API_KEY="{API_KEY}"
claude
$env:ANTHROPIC_BASE_URL="{BASE_URL}"
$env:ANTHROPIC_API_KEY="{API_KEY}"
claude
OpenAI Codex & Python SDK OpenAI Spec
Connect using the official openai Python package or Codex clients. Directly query models like codex, claude-5-opus, or claude-5.5-opus.
from openai import OpenAI
# DirectRouter OpenAI-compatible client
client = OpenAI(
base_url="{BASE_URL}/v1",
api_key="{API_KEY}"
)
response = client.chat.completions.create(
model="codex", # or claude-5-opus, claude-5.5-opus, gemini-3.7-flash
messages=[
{"role": "system", "content": "You are a master software architect."},
{"role": "user", "content": "Write an asynchronous worker queue in Python with retry backoff."}
]
)
print(response.choices[0].message.content)
Cursor IDE Integration
Supercharge Cursor with DirectRouter's low-latency proxy. Follow these quick steps:
claude-5-opus, claude-5.5-opus, codex, gemini-3.7-flash.Aider AI Pair Programmer
Aider works out of the box with DirectRouter's OpenAI compatibility layer:
export OPENAI_API_BASE="{BASE_URL}/v1"
export OPENAI_API_KEY="{API_KEY}"
aider --model claude-5-opus
Responses AI & OpenAI Responses API Modern Agent Standard
Full compatibility with OpenAI's new agentic primitive (/v1/responses and client.responses.create). Supports native tool calling, streaming SSE events, and auto-state management.
client.responses.create)from openai import OpenAI
# DirectRouter Responses API Client
client = OpenAI(
base_url="{BASE_URL}/v1",
api_key="{API_KEY}"
)
# Call via client.responses.create (GPT 6 Astra, Claude 5.5 Opus, etc.)
response = client.responses.create(
model="gpt-6-astra", # or responses-ai, claude-5.5-opus, gemini-3.7-flash
instructions="You are a helpful assistant.",
input="Provide a 2-sentence summary of gravitational lensing."
)
print(response.output_text)
POST /v1/responses)curl -X POST "{BASE_URL}/v1/responses" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer {API_KEY}" \
-d '{
"model": "gpt-6-astra",
"instructions": "You are a helpful assistant.",
"input": "Explain quantum superposition in 1 sentence."
}'
Cline — VS Code Extension OpenAI-Compatible
Cline is an autonomous AI coding agent for VS Code. Connect it to DirectRouter using the "OpenAI Compatible" provider. Supports all models including Claude 5 Opus and GPT 6 Astra.
API Provider: OpenAI Compatible
Base URL: {BASE_URL}/v1
API Key: {API_KEY}
Model: claude-5-opus
RooCode — VS Code Extension Cline Fork
RooCode (formerly Roo-Cline) is a community fork of Cline with enhanced capabilities. Setup is identical — use "OpenAI Compatible" provider with DirectRouter.
API Provider: OpenAI Compatible
Base URL: {BASE_URL}/v1
API Key: {API_KEY}
Model: claude-5-opus
settings.json alternative (manual){
"roo-cline.apiProvider": "openai",
"roo-cline.openAiBaseUrl": "{BASE_URL}/v1",
"roo-cline.openAiApiKey": "{API_KEY}",
"roo-cline.openAiModelId": "claude-5-opus"
}
OpenCode — Terminal AI Agent CLI / TUI
OpenCode is a powerful terminal-based AI coding agent. Configure it to use DirectRouter as its OpenAI-compatible backend via ~/.config/opencode/config.json.
~/.config/opencode/config.json{
"providers": {
"openai": {
"apiKey": "{API_KEY}",
"baseURL": "{BASE_URL}/v1"
}
},
"model": "openai/claude-5-opus"
}
Continue — VS Code & JetBrains IDE Plugin
Continue is an open-source AI code assistant for VS Code and JetBrains. Use DirectRouter as an OpenAI-compatible provider via ~/.continue/config.json.
~/.continue/config.json{
"models": [
{
"title": "DirectRouter — Claude 5 Opus",
"provider": "openai",
"model": "claude-5-opus",
"apiKey": "{API_KEY}",
"apiBase": "{BASE_URL}/v1"
},
{
"title": "DirectRouter — GPT 6 Astra",
"provider": "openai",
"model": "gpt-6-astra",
"apiKey": "{API_KEY}",
"apiBase": "{BASE_URL}/v1"
}
]
}
QwenCoder & Qwen Code OpenAI API Compat
Alibaba's Qwen Code (and the broader QwenCoder family) uses the standard OpenAI Python SDK and env var conventions. Point OPENAI_BASE_URL to DirectRouter and use any available model.
export OPENAI_BASE_URL="{BASE_URL}/v1"
export OPENAI_API_KEY="{API_KEY}"
# Run Qwen Code CLI
qwen-code
$env:OPENAI_BASE_URL="{BASE_URL}/v1"
$env:OPENAI_API_KEY="{API_KEY}"
qwen-code
from openai import OpenAI
client = OpenAI(
base_url="{BASE_URL}/v1",
api_key="{API_KEY}"
)
response = client.chat.completions.create(
model="claude-5-opus", # or gpt-6-astra, gemini-3.7-flash
messages=[
{"role": "user", "content": "Refactor this Python function to be async."}
]
)
print(response.choices[0].message.content)
Standard cURL / HTTP Requests
Test your endpoint directly from the command line:
curl -X POST "{BASE_URL}/v1/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer {API_KEY}" \
-d '{
"model": "claude-5-opus",
"messages": [
{"role": "user", "content": "Hello DirectRouter!"}
]
}'
Activity & Request Auditing LIVE METERING
Comprehensive edge transaction records, token consumption, and metering.
| Timestamp | Model | Prompt Tokens | Completion Tokens | Cost (USD) |
|---|
Credit Balance & Billing WALLET
Prepaid balance metering based on exact input and output token consumption.
Credits never expire. Top up via vouchers, invite friends ($10.00 each), or complete the daily Linkvertise task ($10.00).
Enter any administrator-issued voucher or community promotional code below to credit balance immediately.
Admin Console & Code Generator
MASTER ACCESSCreate redeem vouchers with any dollar value, inspect real-time voucher redemptions, manage network credits, and bypass all IP and rate constraints.
Mint New Redeem Code
Specify any credit amount and maximum allowed redemptions
Leave empty to auto-generate a secure code.
Active Redeem Codes
All redeemable codes in the sovereign database
| Code | Credit Value | Usage (Uses / Max) | Status | Created Date | Actions |
|---|---|---|---|---|---|
| Loading voucher codes... | |||||
Registered Users Directory
Manage balances, inspect IP locks, and adjust quotas
| User | Balance | Registered IP | Invites | Created Date | Actions | |
|---|---|---|---|---|---|---|
| Loading user directory... | ||||||
Live Traffic & Request Stream
Real-time edge completions and metering records
| User | Model | Prompt Tokens | Completion Tokens | Cost (USD) | Latency | Timestamp |
|---|---|---|---|---|---|---|
| Loading live traffic audit... | ||||||
Live Request Logs & Prompt Inspector RING BUFFER (2,000 MAX)
Inspect incoming user prompts, model outputs, and function tool calls with zero-cost rolling retention.
| User / IP | Model & Endpoint | Prompt Snippet | Response / Tool Calls | Tokens | Cost | Latency / Time | Inspect |
|---|---|---|---|---|---|---|---|
| Loading live request logs... | |||||||