Instant · BridgeheadBridgehead
Your account →

The fastest inference × Claude · ChatGPT · any MCP app

Ask. It's already there.
The fastest inference, on your own key.

Cerebras · gpt-oss-120b · Qwen · 3,000 tokens/s · Streaming · Instant viewer · Any screen · Your key · App · Bridge

Join the waitlist

Invite-only beta: join the waitlist and your place is kept. When it opens for you, it simply works in your chat.

Chat answers stream in at reading speed. When you want code, a rewrite or a whole drawing, you sit and watch it type, and the fastest chips in the world aren't in your chat. Instant · Bridge is a connector you add once. It runs the fastest inference providers on your own key, Cerebras first, from the chat you're in: a page of code or a full SVG in a second or two, with how fast it was. It's the first connector designed to work with an app: open the Instant viewer on any screen and the answer appears there as it's written, while your chat catches up, and your assistant still gets every word.

THE USUAL WAY FAST CHIPS, SLOW CHATS INSTANT · BRIDGE Claude / Claude CodeChatGPT / Codex Instant · Bridge Claude / Claude CodeChatGPT / Codex INSTANT · BRIDGE Instant · Bridge ONE ENGINE · FAST PROVIDERS AS PLUG-INS · YOUR KEY

What it can do

Answers in a blinkgpt-oss-120b on Cerebras at about 3,000 tokens a second; the first words in about half a second. Drafts, rewrites and code feel instant.
Drawings and views in a second or twoA full SVG or a self-contained HTML view, written fast enough to feel like it appeared.
Pictures in, answers outQwen on Cerebras reads images: ask about a screenshot or a photo.
Use it with the Instant viewerPick a provider and a model, ask, and the answer appears at once on any screen, a TV, an iPad or a phone, or inline in your chat, while your chat catches up. Your assistant still gets the same answer: one run, never two.
Build your own appableThe viewer is an App · Bridge app. Make your own with App · Bridge: any app can ask Instant · Bridge on the person's own key, and the key never reaches the page.
More providers, one by oneOne engine with providers as plug-ins, each on your own key. Cerebras is first; Groq is next.

Invite-only, for now.

Join the waitlist

01
Join the waitlistSign in once; your place is kept
02
We let you inIn the order people joined
03
Add it to your appYour waitlist page then shows the address

Instant · Bridge is in an invite-only beta. Join the waitlist and your place is kept; when it opens for you, it works in Claude, ChatGPT, Codex and any MCP app.

Join the waitlistAlready in? Your account

Questions

What can I do with Instant · Bridge?

A connector that runs the fastest inference providers in any chat, on your own key. Cerebras is first: gpt-oss-120b at about 3,000 tokens a second, and Qwen, which also reads pictures. Every answer comes back with how fast it was.

What is the Instant viewer?

An app you open on any screen. Pick a provider and a model and ask, from the viewer or from your chat, and the answer appears on the screen as it's written while your chat catches up. Your assistant gets the same answer and can carry on with it. Instant · Bridge is the first connector designed to work with an app like this.

Which key does it use, and what does it cost?

Your own Cerebras key. You pay Cerebras directly, at its prices (gpt-oss-120b is about $0.35 in and $0.75 out per million tokens), and Cerebras has a free tier. The key stays on the server; no page ever sees it.

Will it have other providers?

Yes, one by one: each fast provider plugs into the same engine, on your own key. Groq is next.

Can I use it now?

It's in an invite-only beta: join the waitlist and your place is kept. When it opens for you, it simply works in your chat.