Multi-Model Token Architecture
Modern development teams rarely restrict their workflows to a single model provider. Coding assistants rely on Claude 3.5 Sonnet, cost-sensitive summarization uses DeepSeek V3 or Llama 3.3, and multi-step agentic planning leverages DeepSeek R1 or OpenAI o1.
BlackShield AI Gateway serves as an intelligent, secure reverse proxy. Tenants connect their own API keys (BYOK) for OpenRouter or individual vendors, while BlackShield enforces request sanitization, secret exfiltration defense, and real-time usage telemetry before traffic leaves the tenant perimeter.
OpenRouter Aggregation
Access 200+ models using a single managed API key, complete with attribution headers and automatic fallback routing.
Direct Frontier BYOK
Connect native endpoints for DeepSeek, Together AI, OpenAI, Anthropic, Google Gemini, Groq, and Perplexity.
Live Unmocked Probes
Validate token validity, model responsiveness, and measured latency directly in the console with zero fake telemetry.
Supported Token Vendors
| Vendor | Default Endpoint | Popular Models | Attribution |
|---|---|---|---|
| OpenRouter | https://openrouter.ai/api/v1 | anthropic/claude-sonnet-4.5, deepseek/deepseek-r1 | HTTP-Referer, X-Title |
| DeepSeek | https://api.deepseek.com | deepseek-chat, deepseek-reasoner | Bearer Auth |
| Together AI | https://api.together.xyz/v1 | meta-llama/Llama-3.3-70B-Instruct-Turbo | Bearer Auth |
| Perplexity AI | https://api.perplexity.ai | sonar, sonar-reasoning, sonar-pro | Bearer Auth |
Connecting Your Client Applications
Point your IDE (Cursor, Windsurf, VS Code Continue) or AI agent framework to your tenant AI Gateway proxy. All incoming requests are validated against active guardrails before forwarding to OpenRouter or the specified vendor.
curl -X POST https://app.blackshield.io/api/v1/ai-gateway/openai/v1/chat/completions \
-H "Authorization: Bearer <tenant_gateway_token>" \
-H "Content-Type: application/json" \
-d '{
"provider": "openrouter",
"model": "anthropic/claude-sonnet-4.5",
"messages": [
{ "role": "user", "content": "Analyze AWS IAM trust policy for wildcard principals." }
]
}'