How is this different from OpenRouter?

OpenRouter is a routing API for developers. idapt is a full product (chat UI, files, agents, sharing, billing) with multi-model routing baked in. You also get cheapest-first cascade and sticky pinning for free.

Do I need separate API keys for each model?

No. One idapt subscription gives you usage credit across every supported model, through one key.

Can I see how much each model costs me?

Yes. Every reply shows the routed model and the cost. Settings → Usage breaks it down by model, day, agent, and chat.

What happens when a model is down?

The cheapest-first cascade routes the message to the next equivalent model in the family. You see which model actually answered in the reply metadata.

Chat

One chat. Every model.

GPT, Claude, Gemini, Grok, Llama: pick mid-conversation, cheapest-first.

Use for Free See pricing

We pick the right one. You can override.

Don't want to think about which model? We'll use the cheapest one that handles your question well. Pin a favourite and we'll always use that instead.

One workspace, not four tabs.

Four subscriptions

Four apps. Four histories.

One model per app, stuck in its window.
Conversations scattered across tools.
Upload the same file to each one.
Switch model → start the chat over.

idapt

One subscription. Every model. Same chat.

One bill that covers all of them.
Swap model in the middle of a thread.
Your files are there for every conversation.
Ask two models the same thing, side by side.

Second opinions, one click.

Retry any answer with a different model, right in the same thread. The new answer lands on the same turn, so you compare without retyping.

Try it: pick a model, compare the answers.

One prompt, four frontier models. Click a model to compare the answers: no sign-up, no typing.

Click a model to compare its answer — same prompt, four frontier models.

Best launch tagline for a group-travel app? Punchy, no jargon.

Plan less. Wander more.

One shared itinerary — every traveler on the same page, in real time.

The names you'd expect. And many more.

Every major lab, plus the open-source models worth knowing about. All in the same chat box.

OpenAI

Anthropic

Google

xAI

What you get

200+ models, one chat

OpenAI, Anthropic, Google, xAI, Meta, Mistral, DeepSeek, Qwen, and more: same composer.

Switch mid-thread

Drop GPT for Claude on the next message. The whole context follows.

Side-by-side compare

Send one prompt to two models. Read both replies in parallel tabs.

Cheapest-first routing

Pinned model unavailable? idapt cascades to the next cheapest equivalent automatically.

Sticky pinning

Lock a model per conversation so the cascade never surprises you.

No provider lock-in

Your chat history, files, agents: they all stay with you, not your model vendor.

How it works

1
Type
Open any chat. The model picker is right there next to the composer.
2
Pick or pin
Choose a model for this turn, or pin one so every reply uses it.
3
Switch any time
Change models mid-conversation. The full context comes along.
4
Compare when it matters
Open a parallel tab to send the same prompt to a second model.

Frequently asked

Try the model that fits the question.

Open the chat. Type. We'll pick a good model, or you can choose.

Start a chat Read the docs

Read the docs

Pair with

What you get

200+ models, one chat

OpenAI, Anthropic, Google, xAI, Meta, Mistral, DeepSeek, Qwen, and more: same composer.

Switch mid-thread

Drop GPT for Claude on the next message. The whole context follows.

Side-by-side compare

Send one prompt to two models. Read both replies in parallel tabs.

Cheapest-first routing

Pinned model unavailable? idapt cascades to the next cheapest equivalent automatically.

Sticky pinning

Lock a model per conversation so the cascade never surprises you.

No provider lock-in

Your chat history, files, agents: they all stay with you, not your model vendor.

How it works

Type

Open any chat. The model picker is right there next to the composer.

Pick or pin

Choose a model for this turn, or pin one so every reply uses it.

Switch any time

Change models mid-conversation. The full context comes along.

Compare when it matters

Open a parallel tab to send the same prompt to a second model.