14 frontier models behind a single OpenAI-compatible API. Multi-model failover, native SSE streaming, and transparent per-token metering with zero commitments.
export OPENAI_BASE_URL="https://router.const.web.id/v1"OpenAI-compatible • Anthropic Messages spec • Pay per token • Explore the docs →
Architected specifically for autonomous coding agents, developer tooling, and production workflows that cannot afford dropped connections or unpredictable billing.
Construct custom multi-model failover chains. If a provider returns 429 rate limits, 503 capacity errors, or latency spikes, Const Router automatically routes to the next healthy model without dropping your client connection.
Drop-in replacement for OpenAI and Anthropic SDKs. Keep your codebase, change only the base URL. Every model in the catalog answers to a single const_live_ key.
base_url="https://router.const.web.id/v1"baseURL: "https://router.const.web.id/v1"ANTHROPIC_BASE_URL="https://router.const.web.id"OpenAI Compatible EndpointDirect edge stream passthrough without buffering delays. Exact token usage reports delivered reliably in the stream terminator chunk. Stop the stream, stop paying.
Create unlimited developer keys for team members or autonomous agent loops. Configure individual daily spend caps and allowed model lists to prevent runaway costs.
Every request logged with prompt tokens, completion tokens, latency, and exact micro-cost. Filter by model or time window, inspect raw payload anytime.
Connect any tool, SDK, or coding agent in under 60 seconds by pointing the base URL to Const Router.
export OPENAI_BASE_URL="https://router.const.web.id/v1"
export OPENAI_API_KEY="const_live_xxxxxxxxxxxx"
curl $OPENAI_BASE_URL/chat/completions \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4.1-flash",
"messages": [{"role": "user", "content": "Explain Virtual Combos in 2 sentences."}]
}'https://router.const.web.id/v1Universal OpenAI-compatible host for chat completions and transcription.
https://router.const.web.idDirect Anthropic Messages spec passthrough for Claude Code and Anthropic clients.
X-Const-CostExact micro-cost deducted in USD and latency returned in every response header.
14 frontier models with transparent per-token pricing and instant routing.
| Model | Context Window | Input / 1M Tokens | Output / 1M Tokens | You Save |
|---|---|---|---|---|
deepseek-v4.1-flashDeepSeek v4.1 Flash | 1M | $0.0125 | $0.060 | Save 91.7% |
qwen-3.8-omni-flashQwen 3.8 Omni Flash | 1M | $0.0125 | $0.055 | Save 91.7% |
glm-5.3-flashGLM 5.3 Flash | 1M | $0.0125 | $0.055 | Save 91.7% |
gemini-3.8-flashGemini 3.8 Flash | 1M | $0.020 | $0.080 | Save 97.3% |
glm-5.3GLM 5.3 | 1M | $0.035 | $0.110 | Save 97.5% |
minimax-m3.1-flashMiniMax M3.1 Flash | 1M | $0.035 | $0.135 | Save 76.7% |
qwen-3.8-maxQwen 3.8 Max | 1M | $0.050 | $0.150 | Save 97.5% |
gpt-6.1-solGPT 6.1 Sol | 500K | $0.240 | $1.200 | Save 88.0% |
gpt-6-astraGPT 6 Astra | 272K | $0.450 | $2.200 | Save 95.5% |
claude-haiku-5.5Claude Haiku 5.5 | 1M | $0.020 | $0.100 | Save 80.0% |
claude-sonnet-5.5Claude Sonnet 5.5 | 1M | $0.790 | $3.900 | Save 60.5% |
grok-4.7Grok 4.7 | 500K | $0.140 | $0.450 | Save 93.0% |
Pay strictly for tokens consumed. No subscriptions, no tiers, no recurring card lock-in.
Deposit credits using QRIS or USDC on Solana. Balance drains purely per token consumed down to fractions of a cent.
Engineered for long-running autonomous agent loops, multi-agent workspaces, and high-concurrency coding studios.
What does $1 get you? With deepseek-v4.1-flash, $1 routes over 80,000,000 tokens — enough to analyze an entire large codebase dozens of times.
Start calling frontier models in four straightforward steps.
Sign in with your email or Const account. Nothing else required to explore the catalog.
Deposit USDC on Solana or top up with QRIS. Balance drains per token, never per month.
Create a const_live_ key, configure daily spend ceilings and allowed models.
Point any OpenAI or Anthropic tool to router.const.web.id/v1.
Metering comes directly from upstream provider usage reports, so the number you are charged reflects the exact tokens returned by the model — never a character estimate.
Deposit USDC on Solana, or top up with QRIS (instant IDR)
Input and output tokens counted separately at transparent rates
Cached input tokens billed at up to 90% discount
cost = ((in − cached) × input_price + cached × input_price × 0.10 + out × output_price) / 1M
Balance deducted the moment a request settles
A failed upstream request is never billed
Empty balance blocks new calls, never overdraws or charges a card
Create a key, configure your daily ceilings, and connect your favorite coding agent.