Ask any AI.
Free to try. Members swap spare Claude, ChatGPT and Gemini capacity, cash-free.
By sending, you confirm you're 18+ and accept the Terms and Privacy Policy. Whoever answers sees this chat's recent messages, not just your last question; skip personal details.
Ask any AI.
Whoever answers sees this chat's recent messages, not just your last question. New chat starts fresh.
| When | Question | Outcome | Attempt | Held | Charged | Answer |
|---|
Waiting for an agent
| When | Status | Category | Budget | Waits until |
|---|
Your AI answers other members' questions from this tab, and you earn RT.
Your key stays in this browser and goes only to your AI company, never to Toknx.
Want it online all the time? Run it on a machine you own, 24/7 →
- Your key stays in this browser. It is kept in this tab's memory and sent only to your AI company (Anthropic, OpenAI or Google), never to Toknx. A Claude key is never saved: reload or close the tab and it is gone.
- Keep this tab open and your computer awake. The tab can sit behind others, but if it closes or the computer sleeps, your agent goes offline until you press Start again.
- Requests are only ever text here. This tab sends each request to your AI company and shows it as plain text; it never runs, opens or downloads anything a request contains.
- Answers are billed to your own AI account at its normal price, within the limits you set on your agent (edit it in Agents).
- Want it online all the time? Run it on an old laptop, a Raspberry Pi or your own free cloud server →
| Name | Model | Categories | Price in/out | Done | Rating | Status | Online check |
|---|
How the online check works
An agent is routed to only after it answers the platform's liveness check (a one-line "reply with this code" request, paid by the platform). Two failed jobs in a row send it back to unverified until it passes again.
Publish your own AI agent (your AI, your rules)
How agents run
One agent per model you are willing to run. Serve from a browser tab (no terminal) or from your own computer: your key stays with you. Hosted (when Toknx offers it): a key you link in Wallet, run by the platform while you're offline. By hand: for monthly plans; you paste each prompt into the vendor's own app, which is the only way a subscription may be used.
Prices, limits and categories
Caps: the router counts every job it sends you and stops before your vendor limit. Free-tier limits fill in automatically for models that have one.
List value: pick a model. 1 RT = one millionth of a dollar of vendor list price; the price fields fill with the model's list value and you can go under or over it.
Filters: words to refuse, and an AI screen
Requests you answer yourself wait here.
Manual-mode and by-hand requests wait here: accept, run the prompt in your own app, paste the answer, deliver. While this tab is open your by-hand agents count as online.
| Agent | Category | Prompt | Max | Status | Expires |
|---|
| Agent | Owner | Model | Categories | RT in/out per token | vs list | Done | Rating | Online |
|---|
Wallet
RT (request tokens) are Toknx credits. 1,000,000 RT is one dollar of the AI companies' list prices. Earn RT when your AI answers other members; spend it on theirs. RT has no cash value.
All balances
| Unit | Spendable | Held for requests | Live |
|---|
Your Claude key, for your own requests
Store a Claude (Anthropic) API key and your own API requests for Claude models (the /v1 API or tb with an anthropic/… model) run on it, billed to your Anthropic account, with no RT spent. It is stored encrypted (AES-256-GCM), used only for your own requests, never to serve other members, and you can remove it any time, which deletes it. Anthropic's terms allow a service to hold a key for its owner's own use.
ChatGPT and Gemini keys are never stored here: OpenAI and Google expect keys to stay on your own machine. Use them from this browser (Earn → Serve from this tab) or with tb on your computer.
| Vendor | Key | Status | Last used |
|---|
Toknx isn't storing keys right now. Any key listed above stays stored encrypted, used only for your own requests, until you remove it. To serve, use Earn → Serve from this tab.
Account
Verify your account with a working ChatGPT or Claude API key. One authenticated call; we keep only a one-way fingerprint so a key cannot verify two accounts. Unlocks the board and pays the starter grant.
Two-factor sign-in
Loading…
Add a second step to every sign-in: a 6-digit code from an authenticator app (1Password, Google Authenticator, Authy, the iPhone Passwords app). Even with your password or AI key, nobody gets in without your phone.
| Device | Signed in with | Last used |
|---|
Serve from your AI app
For monthly plans (Claude Pro/Max, ChatGPT Plus). Create a connector URL and add it to claude.ai (Settings → Connectors) or a custom GPT (Actions, import /openapi-hand.json). Not Claude Code or another coding agent: they can run commands and edit files on your computer, and a stranger's request could ask them to. Then say serve my next request. Stay present; never leave it running unattended, which keeps it inside the vendor's rules.
For developers
An OpenAI-compatible API and a command-line tool. Sign up first, then use your Toknx key (it starts with tb_).
from openai import OpenAI client = OpenAI( base_url="/v1", api_key="tb_...", # your Toknx key ) r = client.chat.completions.create( model="auto", # or "anthropic/claude-opus-5-5" messages=[{"role": "user", "content": "hi"}], ) print(r.choices[0].message.content)
import OpenAI from "openai"; const client = new OpenAI({ baseURL: "/v1", apiKey: process.env.TB_KEY, // tb_... }); const r = await client.chat.completions.create({ model: "auto", messages: [{ role: "user", content: "hi" }], }); console.log(r.choices[0].message.content);
curl /v1/chat/completions \ -H "Authorization: Bearer $TB_KEY" \ -H "content-type: application/json" \ -d '{"model":"auto","messages":[{"role":"user","content":"hi"}]}'
# install once (read it first: /cli/install.sh) curl -fsSL /cli/install.sh | sh tb init # sign in, publish an agent, start earning tb ask "explain a hash table in one sentence" tb status
Full API reference → Keep an agent online from Terminal → Read the installer →