Your First Conversation
Send your first message in under 30 seconds.
The 30-second version
- Sign in to app.nyquest.ai
- You're on the dashboard. There's a chat input at the bottom.
- Type a question. Hit Enter.
- The reply streams in word-by-word.
That's it. You're talking to an AI, paying as you go from your wallet (or for free if you've added a BYOK key).
What just happened under the hood
When you hit Enter, several things ran:
- Routing — the platform picked the model you have selected (default: a free fast model from OpenRouter). Your BYOK keys take priority over platform-hosted keys.
- Streaming — tokens stream back one at a time over Server-Sent Events (SSE), which is why you see the reply appear word-by-word instead of all at once.
- Memory recall — if previous conversations contained relevant context, summarized chunks were passed in as additional system context.
- Cost accounting — input + output token counts × the model's per-token rate were debited from your wallet (or set to $0 if you used a BYOK key).
All of this is automatic. Nothing to configure for the basic chat case.
Picking a different model
The model picker is in the header — it shows the currently active model name. Click it to open the picker.
| What you'll see | What it means |
|---|---|
| Model name + provider | e.g. gpt-4o-mini (OpenAI) |
| Per-token cost | input + output rate per million tokens |
| BYOK badge | this provider has your own API key |
| Speed indicator | rough latency tier |
Type to filter. Pick one. Your selection persists across conversations.
Conversations vs messages
A conversation is a list of messages between you and the model. Conversations are saved automatically — you can find them in your conversation list (left sidebar in most chassis).
Each conversation has its own context window. The model sees the full history of that conversation when generating each new reply, up to the model's context limit. If you start a new conversation, the model has no memory of the old one (unless memory recall pulls relevant chunks in).
How long to expect
| Scenario | Approximate latency |
|---|---|
| First-token response from a fast model | 200–600 ms |
| Full short reply (50 words) | 1–2 sec |
| Long reply (1,000+ words) | 5–15 sec |
| Reasoning model (o1, GPT-5) | 10–60+ sec |
Slower replies are normal for reasoning models — they're working through chain-of-thought.
Where to next
- Picking a Model — how to compare and choose
- Streaming — what the live token streaming actually looks like
- BYOK Setup — bring your own API keys to skip platform fees