Every AI model.
One prompt away.
Chat with Claude, Llama, DeepSeek and more in one place. Free, no signup. Compare two answers side by side, then ship the one you like with a single key.
An AMM is a pool holding two tokens, say ETH and USDC.
The price is simply the ratio between them: if the pool holds 10 ETH and 30,000 USDC, one ETH costs about 3,000 USDC.
When you buy ETH, you add USDC and remove ETH, so ETH becomes scarcer in the pool and its price rises.
The pool keeps x × y = k constant, which is why big trades move the price more.
Arbitrage traders keep this price in line with other markets.
The short version
An automated market maker replaces the order book with a pool of two tokens and a formula.
- The pool holds reserves of both tokens, for example ETH and USDC.
- The price is the ratio of those reserves, not a match between buyers and sellers.
- Every trade changes the reserves, so every trade changes the price.
- The rule
x × y = kkeeps the product of reserves fixed, which makes larger trades cost more per unit (slippage). - When the pool drifts from the wider market, arbitrageurs trade it back and pocket the gap.
The result: prices update continuously, with no one quoting them.
One catalog for the models teams actually use
- Anthropic
- OpenAI
- Meta
- Mistral
- DeepSeek
- Qwen
- xAI
- Cohere
Built for choosing, then shipping.
Everything you need to find the right model and put it to work, without a single signup form.
See speed and quality in the same view
Every run shows time to first word, total time and length for each model, next to the answer itself.
Your wallet is your account
No email, no card. Sign once, create keys, revoke them any time.
A one-line migration
Dualyne uses the OpenAI format. Change the base URL and your app, bot or agent keeps working.
Free access, paid for in public
Trading fees fill a treasury that covers everyone's free requests. Every dollar in and out is on the ledger.
Every leading model, one endpoint.
Browse the catalog, or see how models rank when people compare them head to head.
| Model | Provider | Best for | Speed | Context | In / out per 1M | Tier | Status |
|---|---|---|---|---|---|---|---|
| Claude SwiftClaude Haiku 4.5 | Anthropic | Quick answers, chat, summaries | 200K | $1.00 / $5.00 | Explorer | Live | |
| Claude BalancedClaude Sonnet 4.5 | Anthropic | Writing, analysis, everyday coding | 1M | $3.00 / $15 | Holder | Live | |
| Claude DeepClaude Opus 4.5 | Anthropic | Hard reasoning, long documents | 200K | $5.00 / $25 | Holder | Live | |
| GPTGPT-5 | OpenAI | General purpose, tool use | 400K | $1.25 / $10 | Holder | Live | |
| GeminiGemini 2.5 Pro | Long context, images | 1M | $1.25 / $10 | Holder | Live | ||
| LlamaLlama 3.3 70B Instruct | Meta | Open weights, low cost | 131K | $0.10 / $0.32 | Explorer | Live | |
| DeepSeekDeepSeek V3.1 | DeepSeek | Math, code, budget reasoning | 164K | $0.25 / $0.95 | Explorer | Live | |
| MistralMistral Small 3.2 24B | Mistral | Fast multilingual, low cost | 256K | $0.09 / $0.25 | Explorer | Live |
Prices are what the provider charges per million input / output tokens. Explorer and Holder requests are free, paid for by the treasury; Builder pays cost + 15%.
Run a comparison
One key. The whole catalog.
Switch models by changing one string. Keep the SDKs and tools you already use.
- Streaming, tools and JSON mode pass straight through.
- Usage per key, so you know what each app spends.
- Pay per request in USDG or ETH once you outgrow the free tier.
curl https://api.dualyne.com/v1/chat/completions \ -H "Authorization: Bearer $DUALYNE_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-swift", "messages": [{"role": "user", "content": "Hello"}], "stream": true }'
Pricing that grows with you.
Every wallet gets a daily allowance. Holding the token raises it. Past that, pay only for what you use.
For anyone with a wallet who wants to try the API.
- 20 requests a day
- Fast models
- 1 API key
For wallets holding the token. Paid for by the treasury.
- 250 requests a day
- Every model in the catalog
- 5 API keys
- Priority routing
For apps in production. Top up in USDG or ETH.
- No daily cap
- Every model in the catalog
- Unlimited API keys
- Usage reports per key
Limits shown are the launch proposal and may change before the token goes live.
The token that pays for the compute.
Every trade adds to the treasury that funds free access. Holding it raises your limits.
$DLYN
Dualyne access token- Total supply
- 1,000,000,000
- Trade fee
- 1% to the treasury
- Team allocation
- None at launch
- Holder tier
- 100,000 DLYN
Where each fee goes
Proposal$DLYN is a utility token that raises your usage limits on Dualyne. It is not an investment, share or promise of profit, and its price can fall to zero. Check the rules where you live before buying. See the Terms.
Treasury
Every fee in and every request out, published daily.
Sample data| Day | Fees in | Inference |
|---|---|---|
| Sep 22 | +1,284 | −452 |
| Sep 21 | +1,902 | −431 |
| Sep 20 | +2,650 | −388 |
| Sep 19 | +3,118 | −341 |
| Sep 18 | +4,407 | −296 |
Questions, answered.
Is Dualyne really free?
Yes. Comparing models in the browser is free with no account. Wallets get a daily allowance through the API, paid for by the treasury.
Why do I need a wallet for API keys?
Your wallet works as your account, so there's no email or password to leak. It also lets us raise your limits automatically when you hold the token.
Which models can I use?
The API routes to models from Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek and more. Every model in the catalog is live, in the Compare tool and through the API.
What happens to my prompts?
Dualyne does not keep the text of your prompts or the answers. After a comparison, they stay in memory for up to an hour so you can choose to share them, and are saved only if you click Share. We keep only what's needed to run limits and publish the ledger: which model you used, token counts, cost and timing. Requests pass through OpenRouter to the model's provider, whose own data policy applies. Details are in the Privacy Policy.
What happens when my daily allowance runs out?
Your requests pause until the next day, or you can top up credits in USDG or ETH and keep going at model cost plus 15%.
Where does the treasury money come from?
A 1% fee on every token trade goes to a public treasury wallet. The ledger shows what comes in and what is spent on inference, every day.
Why not call OpenRouter or the providers directly?
You can, and for heavy paid use it's cheaper. Dualyne adds what they don't: a free daily allowance paid for by the treasury, a side-by-side Compare tool, and keys tied to a wallet instead of an email and a card.
Can I use Dualyne with my existing code?
Yes. Dualyne speaks the OpenAI API format. Change the base URL and the API key, and existing SDKs, bots and agents keep working.