Standard inference
Choose models, provider preferences and budgets. Health checks, price and latency guide the route and its fallbacks.
Route AI calls through one API, one prepaid balance and a signed receipt for every generation. Choose your model. Keep control of the route.
Encrypted chat, an agent rulebook and a network open for early hosts. Built on Solana.
Try every model and its toolsFind your model by capability, context and price.
Every model and its tools, in one place.
Keep your SDK. Change two lines.
Inspect the proof behind a generation.
Discover opt-in agents and their track records.
An open network for early hosts.
Budget caps per request, hour, day and week.
Agent rulebookSingle-use ask-first approval window on the official Agents page.
Owner approvalDeposit required to host. Admission by hardware attestation and probation.
Network hostsValidity of a router-signed track-record certificate.
CertificatesRouting, payments, proof and privacy in one layer. Keep your SDK; get a receipt, a budget and a choice of how to pay.
Any model, any provider, any route. One familiar API for your next generation of agents. Know where every call goes and what it costs, down to the receipt.
Keep the clients and request shapes your team already knows. Change the base URL and key, with OpenRouter-compatible routing preferences built in.
One balance for every model and provider. Fund it with supported assets from your own wallet.
Coming soonSee the model, provider, tokens and cost behind each generation. A signed receipt travels with the response, and anyone can inspect it.
Receive a 402 quote, pay, then retry with a payment proof. Follow the documented flow for eligible agent requests.
Cap spending per request, hour, day and week. Choose models, lanes, tools and working hours. Ask first, or stop the agent.
Opt in to public profiles with random slugs, owner-chosen capabilities and a verifiable track record.
Publish a sidecar with the sealed recipe. The router checks a registered agent’s TDX quote. No sealed agent is registered on the official site yet.
Link Telegram with a one-time code. Approve or deny single-use requests and receive alerts. Approval details pass through Telegram.
Client-side encryption through the attested gateway. The router forwards ciphertext; verification and device setup are part of the trust model.
Join with an approved confidential build. Hardware attestation, host policy and probation determine admission.
Supported token routes use daily caps, oracle-based valuation and recorded allocations. Settlement is accounted for in SOL.
Provider health, price, latency and your policy guide fallback selection when a preferred provider is unavailable.
Inspect usage records and integrate routing telemetry with the observability tools your team already uses.
Verifiable receipts, ledger exports and router-signed track-record certificates. Inspect the scope and lifetime of each claim.
Workspace roles, shared access and individual budgets keep team usage accountable and controlled.
Coming soonThe request shape stays familiar. Choose a route that fits your model, payment preferences and data policy.
Choose models, provider preferences and budgets. Health checks, price and latency guide the route and its fallbacks.
Confidential computing evidence and policy checks establish the conditions for a private route. Inspect the verification requirements.
Pay through supported token routes. Daily caps, oracle valuation and recorded allocations keep settlement accountable.
Encrypt on your device. Ciphertext is forwarded through the attested gateway with device setup and verification requirements.
A familiar request goes in. An accountable route comes back.
Use an existing API request format. Specify the model, provider preferences and budget.
The router evaluates provider health, pricing, data policy and request requirements.
Receive usage information, itemized costs and a signed receipt that can be independently checked.
Approved builds. Fresh hardware evidence. A signed policy. Admission is something you can inspect.
The open host network begins with confidential computing. Hosts submit hardware attestation and policy evidence before admission and probation.
Public records make approved builds, operator checks and recorded evidence inspectable. The trust boundary is explicit, not implied.
Host payouts and host-bond slashing are not switched on. No network host payouts are being made.
Approved build
Fresh hardware quote
Signed host policy
Operator and payout screening
Admission and probation
Keep your workflow. Choose the controls. Every call stays within the rules you set.
Keep existing SDKs, provider choices and routing preferences.
Set request, hourly, daily and weekly spending limits.
Use approval windows and inspect signed receipt ledgers.
Alerts, circuit breakers and progressive autonomy keep agents accountable.
Model, provider, payment and proof. Together in one record.
An agent selects a model and a supported Stock Token as its payment asset. The route checks its budget and daily cap, then records token usage and settlement accounting.
The receipt itemizes cost in SOL and records the selected payment asset. A signature can be checked against the documented trust model.
This example is not a live transaction. No payment was made, signature verified or network anchor submitted.
Read the case studyWhat is available. What comes next. And what is deliberately switched off.
Device encryption and the attested gateway privacy protocol.
Budgets, model and tool controls, Telegram approvals and alerts.
Early hosts, approved builds, hardware attestation and probation.
File Q&A and documented payments on the official product. Agent wallets, GPU hosts and $ANYR burns are upcoming.
Network host payouts and host-bond slashing are not enabled.
Anyroute is an open inference router for Solana. A familiar API is the starting point. Clear payments, provider accountability and portable receipts are the reason to go further.
Routing, SOL settlement, signed receipts and attestation checks put an inspectable record behind each request.
“A route should tell you more than whether a request succeeded.”
Keep the OpenAI-compatible SDK you already use. Point it at Anyroute and choose your model.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: process.env.ANYROUTE_BASE_URL,
apiKey: process.env.ANYROUTE_API_KEY,
});
const response = await client.chat.completions.create({
model: "meta-llama/llama-3.3-70b-instruct",
messages: [
{ role: "user", content: "Hello, Anyroute." }
],
});
console.log(response);One key, one prepaid balance and a receipt for every call. Change two lines and keep your SDK.