Isolated Private Inference
Local AI models run inside a dedicated firewall-locked KVM node with zero external data leakage.
Run private local models on isolated hardware with drop-in OpenAI and Anthropic endpoints, atomic credit reservations, and full bilingual control.
Local AI models run inside a dedicated firewall-locked KVM node with zero external data leakage.
Every request uses two-phase hold and settlement arithmetic with an immutable append-only audit trail.
Use /v1/chat/completions and /v1/messages with streaming SSE and standard SDKs.
Strict server-side RBAC for User, Reseller, Admin, and Superadmin with tenant isolation.
1. Register, verify your email, and activate a plan or top-up credits.
2. Generate a one-time sk-lopi API key with optional spend and rate limits.
3. Call /v1/chat/completions or /v1/messages and track real-time usage and ledger entries.
Live active models currently served by the LOPI gateway
Turbo RAM-resident Qwen 2.5 Coder 1.5B (~20 tok/s instant response, zero swap).
Context: 32,768 tokens
In: $0.05 / Out: $0.15 per 1M
Turbo RAM-resident DeepSeek R1 1.5B reasoning model (fast chain-of-thought, zero swap).
Context: 32,768 tokens
In: $0.08 / Out: $0.24 per 1M
Private local Qwen 2.5 Coder 7B optimized for code generation and structured API responses.
Context: 32,768 tokens
In: $0.10 / Out: $0.30 per 1M
Private local DeepSeek R1 8B reasoning model running inside isolated VM 107.
Context: 32,768 tokens
In: $0.15 / Out: $0.45 per 1M
Predictable USD and IDR pricing configured directly in the platform catalog
Equivalent: IDR 64,000 · 7 days
Credits included: $4.00
Up to 30 RPM · 30000 TPM
Select PlanEquivalent: IDR 160,000 · 30 days
Credits included: $10.00
Up to 60 RPM · 60000 TPM
Select PlanEquivalent: IDR 1,600,000 · 365 days
Credits included: $100.00
Up to 120 RPM · 120000 TPM
Select Plancurl -X POST https://lopi-app.rudd-catla.ts.net/v1/chat/completions \
-H "Authorization: Bearer sk-lopi-demo" \
-H "Content-Type: application/json" \
-d '{"model":"qwen2.5-coder:7b","stream":true,"messages":[{"role":"user","content":"Hello"}]}'
curl -X POST https://lopi-app.rudd-catla.ts.net/v1/messages \
-H "x-api-key: sk-lopi-demo" \
-H "anthropic-version: 2023-06-01" \
-H "Content-Type: application/json" \
-d '{"model":"deepseek-r1:8b","max_tokens":256,"messages":[{"role":"user","content":"Hello"}]}'
Unused reserved credit holds are automatically released back to your available balance without double-charging.
Yes, every page and API error supports full English and Bahasa Indonesia localization.
Yes. Each sk-lopi key supports optional spend caps, per-key RPM/TPM rate limits, expiration dates, and model allowlists.
Local inference runs exclusively on an isolated KVM node protected by strict UFW firewall rules accessible only from the application gateway.