chat
Chat Completions
POST /v1/chat/completionsOpenAI-compatible messages, tools, images, and SSE streams. Drop-in for the usual SDKs.
open signup · 90%–99% off
One wire.
Every model.
90%–99% off.
Create an account, deposit once, and lock your discount for life. Start at 90% off. One endpoint for Chat Completions, Responses, and Anthropic Messages — with metering that never touches your prompts.
starts at 90% off99% at $1,000.000 prompts stored
route / all traffic online
POST/v1/chat/completionschat
POST/v1/responsesresponses
POST/v1/messages
translate → native upstream
metered at the edge0 tokens
successful requests0
tokens metered0
models online5
availability—
deposit tiers / locked for life
The more you put in, the less each token costs.
Your discount is based on lifetime top-ups, not your current balance. Start at 90% off. Deposit $10.00 to lock 95%, up to 99% at $1,000.00. Deposit once, keep the tier.
Start
90% off
$0.30 / 1M on a $3 model
$10 deposit
95% off
$0.15 / 1M on a $3 model
$50 deposit
96% off
$0.12 / 1M on a $3 model
$200 deposit
97% off
$0.09 / 1M on a $3 model
$500 deposit
98% off
$0.06 / 1M on a $3 model
$1,000 deposit
99% off
$0.03 / 1M on a $3 model
Example prices use a $3.00 / 1M-token model. Discounts apply to every model's base input and output price. Deposits are prepaid credit, not cash.
three dialects / one base url
Bring the SDK you already use.
Every provider route has a native dialect. The proxy speaks all three and translates on the way through, including streaming and tool calls.
responses
Responses
POST /v1/responsesInput items, instructions, function calls, and the Responses event stream — translated both ways.
messages
Anthropic Messages
POST /v1/messagesSystem blocks, tool_use/tool_result, stop reasons, and the Messages stream — same key, same host.
the ledger, not your logs
We keep the numbers. Not the conversation.
The proxy has to count tokens, measure latency, and charge the ledger. It does not need to keep what you asked or what the model said. So it doesn't.
Read the privacy contractPrompts stay yours
Request and response bodies are streamed through and never written to disk.
No IP addresses
No access logs, no client IPs, no headers. Authentication is by API key only.
Metered, not mined
Tokens, cost, latency, TTFT, status, and provider. That's the whole row.
live catalog
Out in the open.
Prices per million tokens, 24-hour uptime, throughput, and latency.
| Model | Input / 1M | Output / 1M | Uptime 24h | Tokens/s | Latency | TTFT | Requests | |
|---|---|---|---|---|---|---|---|---|
| DeepSeek v4.1 Flashdeepseek-v4.1-flash | $0.0300$0.3000cache $0.000600 | $0.1200$1.20 | — | — | — | — | 0 | |
| GLM 5.3glm-5.3 | $0.1400$1.40cache $0.0260 | $0.4400$4.40 | — | — | — | — | 0 | |
| GLM 5.3 Flashglm-5.3-flash | $0.0150$0.1500cache $0.003000 | $0.0500$0.5000 | — | — | — | — | 0 | |
| Kimi K3kimi-k3 | $0.3000$3.00cache $0.0300 | $1.50$15.00 | — | — | — | — | 0 | |
| Qwen3.8 27Bqwen3.8-27b | $0.0250$0.2500cache $0.0150 | $0.2000$2.00 | — | — | — | — | 0 |
public signup / no invite
Pick a username. Start at 90% off.
Deposit $10.00 for 95%, up to 99% at $1,000.00. Prepaid credits, transparent metering, and no prompt storage.