Your Inference
Is Your Own.
Autonomous gateway infrastructure
for OpenAI-compatible inference, multi-provider model routing, and deterministic settlement.
OpenAI_SDK_DropIn
Zero refactoring required. Standardized endpoint specifications for /v1/chat/completions and /v1/models across any modern client.
Sovereign_Billing
Pay-as-you-go micro-settlement in IDR. Atomic hold reservation with instant diff reconciliation and zero monthly lock-in.
Telemetry_Node
Real-time token metrics, latency traces, and cryptographically verified audit trails for every inference transaction.
Integration Protocol
curl https://api.koneva.id/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer kg_live_YOUR_API_KEY" \
-d '{
"model": "deepseek-v4-flash",
"messages": [{"role": "user", "content": "Halo Koneva"}],
"stream": false
}'| Model ID | Provider | Context | Input / 1M | Output / 1M |
|---|---|---|---|---|
| glm-5.3-flashx | GLM | 200,000 | Rp 300 | Rp 1.000 |
| qwen-3.8-flash | Qwen | 262,144 | Rp 300 | Rp 1.000 |
| gemini-3.8-flash | 1,048,576 | Rp 400 | Rp 1.200 | |
| glm-5.3-flash | GLM | 1,000,000 | Rp 400 | Rp 1.200 |
| mimo-v2.6-flash | Other | 262,144 | Rp 400 | Rp 1.200 |
| deepseek-v4-flash | DeepSeek | 1,048,576 | Rp 450 | Rp 1.300 |
Institutional
Gateway
Doctrine
By initiating router traffic, microservices acknowledge the deterministic execution, atomic settlement, and zero-prompt-persistence model of the Koneva gateway protocol.
GATEWAY INTEGRITY
Bearer Credential Authority
Access authority is derived solely from the presentation of cryptographically valid kg_live_ bearer tokens. Koneva stores salted cryptographic hashes, never plaintext keys.
Ephemeral Prompt Transport
Payloads and completions are strictly transient. The gateway processes requests through memory streams without persisting user prompts or model completions to disk.
Upstream Sovereign Failover
Health probes continuously monitor latency and error rates across frontier provider endpoints (DeepSeek, Anthropic, OpenAI, Google) with sub-second route protection.
Drop-In Compatibility
Exact structural conformity with the OpenAI SDK. Endpoint specifications for /v1/chat/completions and /v1/models match upstream protocol schemas without custom dependencies.
SETTLEMENT RULES
Atomic Micro-Reservation
When a completion starts, estimated tokens are reserved atomically. Unused hold amounts are automatically and immediately refunded to available balance upon stream termination.
Deterministic Rupiah Math
Inference costs are calculated per token using exact formula: (input_tokens / 1M × input_price) + (output_tokens / 1M × output_price), rounded to nearest Rupiah.
30-Day Extension Cycle
Account balances remain active for 30 days from the most recent top-up event. Any subsequent deposit automatically extends the validity of the entire combined balance.
Instant QRIS & VA Settlement
Top-ups are reconciled via DOKU automated webhook pipelines with instantaneous ledger commitment and real-time wallet notification.
SECURITY ENCLAVE
Sub-Second Key Revocation
Keys marked for deletion are instantaneously invalidated across all edge clusters. Once revoked, requests with the credential receive deterministic 401 Unauthorized.
Rate-Limiting Protection
Configurable RPM boundaries safeguard backend microservices against runaway concurrency loops, billing spikes, and rogue script execution.
Origin Non-Custodial Boundaries
Koneva operates as an API routing and settlement layer. Infrastructure users maintain total control and ownership of their application data and downstream products.
Latency Telemetry Streaming
Every request is traced with unique request IDs, duration milliseconds, prompt/completion token consumption, and timestamped audit logs.
REGION_ORIGIN: JAKARTA_INDONESIA
ORCHESTRATION_ID: KONEVA-ROUTER-CORE-4.0
Operational Inquiries
Bagaimana cara menghubungkan aplikasi ke gateway Koneva?
Gunakan SDK resmi OpenAI seperti biasa. Arahkan baseURL ke endpoint gateway Koneva dan autentikasi dengan API key kg_live_ Anda.
Apakah 100% kompatibel drop-in dengan struktur response OpenAI?
Ya. Endpoint /v1/chat/completions dan /v1/models menggunakan format OpenAI-compatible secara identik tanpa perlu adapter.
Bagaimana sistem billing per token bekerja?
Koneva memakai model prabayar pay-as-you-go murni. Saldo dipotong presisi berdasarkan token input dan output aktual per request tanpa biaya langganan bulanan.
Berapa masa berlaku saldo akun?
Saldo aktif selama 30 hari sejak tanggal pengisian terakhir. Top-up baru dalam nominal berapapun otomatis memperpanjang seluruh saldo Anda.
Official Telegram Channel & Support Center
Bergabung dengan komunitas developer Indonesia, pantau status upstream model AI secara real-time, dan dapatkan asistensi teknis langsung melalui channel Telegram resmi kami.