323 models · from $0.00032 a message

The proxy for
character chat.

One URL and one key for JanitorAI, SillyTavern, Chub and Risu. Claude, DeepSeek, Gemini, Grok, GLM and 164 uncensored builds, priced per message. Crypto or card from $10, no KYC.

Get a proxy keyGet a proxy key (opens in new tab)JanitorAI setup
  • A price per message on every model, from live rates
  • The uncensored builds and the flagships on one URL
  • A signed receipt per message and a public daily model check
Works inJanitorAISillyTavernChubRisuNo app? Chat in the browser at Unrestrictable.
Proxy URLhttps://api.relayswitch.io/v1/chat/completions
Price per messageone 2K-in / 300-out turn, live rates
323

Models on one proxy URL

From 47 developers, each at its live per-token price; 164 of them uncensored builds. The Models page is the catalog, read at render.

47

Developers

312/336

Unchanged on today’s check

85%

Max under list

$10

Monthly plan: $20 of usage

Monthly usage plans

A month of usage on every model. By card a plan renews by default, off at checkout or in Billing; with crypto it never renews.

Plus

$10
/ month
Get Plus

  • $20 of usage a month
  • Every model

Pro

$25
/ month
Get Pro

  • $50 of usage a month
  • Every model

Max

$100
/ month
Get Max

  • $200 of usage a month
  • Every model

Ultra

$250
/ month
Get Ultra

  • $500 of usage a month
  • Every model

What the key gets you.

Every model on one URL, a public record of which model answered, and a budget on each key you hand out.

The key API
Key
openclaw-main
Budget
$20
Allowed models
claude-sonnet-5-5, gpt-6.1-sol, deepseek-v4.1-flash
Requests / min
60
Tokens / min
400,000
Monthly cap
$100

A budget and limits per agent

Each agent runs on its own key: a prepaid budget in USD, an allowed-model list, a Requests / min and tokens / min limit, a monthly cap and an expiry, set in the console or by one POST to the key API with an admin key. At its budget the key answers 402 and nothing else on the account changes.

bash
1curl https://api.relayswitch.io/v1/chat/completions \
2 -H "Authorization: Bearer $RELAYSWITCH_KEY" \
3 -H "Content-Type: application/json" \
4 -d '{"model": "claude-sonnet-5-5",
5 "messages": [{"role": "user", "content": "Hello"}]}'

One key for 323 models, with failover

Set the base URL to https://api.relayswitch.io/v1 once. Claude, GPT, Gemini, DeepSeek, Kimi and GLM answer the same OpenAI-compatible endpoint and the Anthropic Messages format too, and when a host answers 5xx or 429 the request goes to the next host for that model at the same price.

312 / 336Models unchanged today11 changed, 13 not reached; read 2026-11-12T02:57 UTC
bash
1curl https://api.relayswitch.io/v1/requests/REQUEST_ID/receipt \
2 -H "Authorization: Bearer $RELAYSWITCH_API_KEY"

A public daily check and a signed receipt

Every model gets the same canary prompt each day and the result is public at https://api.relayswitch.io/v1/proof. Every request has a signed receipt with the model, the billed tokens and the charge, checked on the Verify page.

Setup

Paste it into your app.

The field each app has, the value it takes, and the model id. Every guide shows the rest of the panel as the app labels it, with a price per message on each model.

All setup guides

JanitorAI

JanitorAI guide
text
1Proxy URL https://api.relayswitch.io/v1/chat/completions
2API key sk-relay-...
3Model glm-5.3-flash-uncensored

SillyTavern

SillyTavern guide
text
1Custom Endpoint (Base URL) https://api.relayswitch.io/v1
2API key sk-relay-...
3Model glm-5.3-flash-uncensored

Chub

Chub guide
text
1OpenAI Reverse Proxy URL https://api.relayswitch.io/v1
2API key sk-relay-...
3Model glm-5.3-flash-uncensored

Risu

Risu guide
text
1URL https://api.relayswitch.io/v1
2API key sk-relay-...
3Model glm-5.3-flash-uncensored

Or run an agent on it.

The same base URL in the file each framework reads. Each guide carries the full block and a session priced live.

All agent setup guides
json
1// ~/.openclaw/openclaw.json
2models: {
3 providers: {
4 relayswitch: {
5 baseUrl: "https://api.relayswitch.io/v1",
6 apiKey: "${RELAYSWITCH_API_KEY}",
7 api: "openai-completions",
8 models: [{ id: "claude-sonnet-5-5" }],
9 },
10 },
11}

OpenClaw

A models.providers block in openclaw.json. OpenClaw guide

bash
1export ANTHROPIC_BASE_URL="https://api.relayswitch.io/v1"
2export ANTHROPIC_AUTH_TOKEN="sk-relay-..."
3claude

Claude Code

Two environment variables; the Claude ids answer the Anthropic Messages format. Claude Code guide

Tools your agents call.

Server tools inside the API call, routing, caching, guardrails, an MCP gateway and traces, on the same key. Every feature

Web search, fetch and code

Web search ($0.01 a search), page fetch and a code sandbox, run by the API inside one request. Agent tools

MCP gateway

Register MCP servers once; every agent gets one endpoint on its own key. MCP gateway

Auto router and fallbacks

relayswitch/auto picks the tier per request; up to 5 fallbacks and saved presets. AI model router · Custom models · Routing

Cache and compression

A repeat answered for a tenth of its charge; middle-out when the history outgrows the window. Caching

Guardrails and sign-in

Blocked patterns, a prompt-injection check, tool allow and deny lists, and app sign-in. Guardrails

Files and redaction

PDFs read by any model; secrets and personal data swapped out before the prompt leaves. Files

Sessions and log export

A timeline per run, export to Langfuse, Datadog or OTLP, session caps and a loop guard. Observability

Embeddings, audio and batch

8 embedding models, rerank, transcription, image edits and the Batch API. Endpoints

Hosted agents and workflows

Agents with skills, MCP, memory and cron, and JSON workflows run by one request. Agents

Live prices

The flagships, priced live

USD per million tokens, read from the catalog at render.

All 323 models
Anthropic
50% under
Claude Fable 5.2claude-fable-5.2
Input / M
$4.00
Output / M
$20.00
Anthropic
51% under
Claude Opus 5.5claude-opus-5-5
Input / M
$2.716
Output / M
$9.70
DeepSeek
3% under
DeepSeek V4.1 Flash 1017deepseek-v4.1-flash-1017
Input / M
$0.2425
Output / M
$1.164
Z.ai
70% under
GLM 5.3 Flashglm-5.3-flash
Input / M
$0.0873
Output / M
$0.1455
OpenAI
60% under
GPT-6 Lunagpt-6-luna
Input / M
$0.0625
Output / M
$0.20
Qwen
60% under
Qwen3.8 Maxqwen3.8-max
Input / M
$0.80
Output / M
$2.40
MiniMax
30% under
Minimax M3.1 Flashminimax-m3.1-flash
Input / M
$0.105
Output / M
$0.42
Google
40% under
Gemini 3.8 Flashgemini-3.8-flash
Input / M
$0.4925
Output / M
$2.25
xAI
70% under
Grok 4.7grok-4.7
Input / M
$0.60
Output / M
$1.80
Moonshot AI
70% under
Kimi K3kimi-k3
Input / M
$0.90
Output / M
$4.50

Built for long runs

Budgets, limits and retries per agent. Prepaid per token by card or crypto, or a monthly usage plan.

A budget per key

A prepaid budget in USD on each key. At the budget the key answers 402; the rest of the account keeps running.

Limits per key

Requests / min, tokens / min, a monthly cap and an expiry, each set on the key in the console or by API.

Allowed models

A list of up to 100 ids a key may call. An agent on that key cannot reach for anything else.

Failover at the same price

A host that answers 5xx or 429 hands the request to the next host for that model. A 429 on your key carries retry-after.

OpenAI and Anthropic formats

Chat Completions, Responses and Anthropic Messages on every chat model. Streaming and tool calls on all of them.

Prompts are relayed, not kept

Bodies are not stored or used for training. Caching and body retention are per-key switches, off until you turn them on.

Wholesale

The upstream for people who resell AI.

With approval, one prepaid key behind your proxy, bot, app or relay station: per-key model lists and monthly caps, a signed receipt on every call, a public daily check that each model is the one on the label.

Wholesale keys
FAQs

Questions

Payment, keys and billing in short.

Open the console

A key in under a minute

Get a key

Or email hello@relayswitch.io

One URL, every model, priced per message

Top up from $10 by crypto or card, create a key, paste https://api.relayswitch.io/v1/chat/completions into your app.

Setup guidesGet a proxy key
Join on Telegram