# Multi-Model Chat

Canonical page: `https://nanthai.tech/features/multi-model-chat`

Tier: Free

Ask one question, get answers from up to three AI models at once.

Send the same prompt to multiple AI models simultaneously. Compare responses, fork conversations, and queue follow-ups while models are still thinking.

## How it works

1. **Pick your models** — Tap the + button in the chat input bar and choose Participants. Browse 150+ models with filters, sort options, and a "Help me choose" wizard that recommends models based on your task.
2. **Send one prompt** — Type your question or instruction once. Edge sends it to all selected models simultaneously and streams their responses in real time.
3. **Compare vertically** — Responses appear vertically stacked in a shared card — each with its own avatar, model label, and action buttons. Scroll through to compare reasoning, style, and accuracy.
4. **Fork or continue** — Found the best answer? Fork that response into its own conversation branch and keep chatting with just that model. Your other branches are always one tap away.

## What you can do

- **Compare up to 3 models at once** — Run the same prompt against different models simultaneously. See how each one interprets your question.
- **Fork into separate branches** — Pick the best response and fork it into a standalone conversation. Navigate between branches with the branch pill.
- **Queue follow-ups while streaming** — Don't wait for models to finish. Type your next message and it'll be sent as soon as all responses are complete.
- **Save favourite model combos** — Pin your go-to model combinations to the favorites strip for one-tap access on any new conversation.
- **Retry with a different model** — Not happy with one response? Tap retry-different to swap that slot to a new model without losing the rest.
- **Mix models and personas** — Combine any model with a persona's system prompt. Compare how the same persona performs on different models.
- **Composes with search, personas, and more** — Multi-model works with web search, personas, memories, and knowledge base attachments — each participant resolves its own context independently.

## Perfect for

- Comparing how different models write code — testing the same algorithm prompt on multiple models to see which produces the cleanest implementation.
- Getting a second opinion on important decisions — asking multiple models to evaluate a business plan, legal question, or technical architecture.
- Finding the best tone for writing — seeing how different models handle creative writing, emails, or marketing copy side by side.
- Learning and research — comparing explanations of complex topics to find the clearest, most accurate answer.
- Testing prompt engineering — seeing how your carefully crafted prompt performs across different model architectures.

## Related features

- Participant Options (`/features/participant-options`)
- Branching (`/features/branching`)
- Personas (`/features/personas`)