01 / signature
Multi-model, one chat.
Swap between GPT-4o, Claude Opus, o4-mini, and the rest without losing context. Run two models side by side when you want a second opinion. One conversation, six minds.
GPT-4oOpus 4.7Sonnet 4.6Haiku 4.5o4-mini
· A thoughtful AI chat
Two minds.One conversation.
Switch between GPT and Claude mid-conversation. Branch to explore alternatives without losing the thread. Attach files, ground with web search, keep every word. Three free messages a day — upgrade for unlimited.
6 models ready3 free / day
Today's chat
Pick any model.
- Claude Opus 4.7Anthropicpro
- GPT-4oOpenAIfree
- Claude Sonnet 4.6Anthropicfree
- o4-miniOpenAIfree
- Claude Haiku 4.5Anthropicfree
- GPT-4.1OpenAIpro
6 models · 2 providersone chat
Powered by the frontier
- OpenAI GPT-4o
- GPT-4.1
- o4-mini
- Anthropic Claude Opus 4.7
- Sonnet 4.6
- Haiku 4.5
· Why quiet
Most AI apps are loud.
We built the other kind.
Onboarding wizards, autoplay demos, robot mascots, banners promising AGI by Thursday. Pop-ups asking if you're really sure you want to leave. We skipped all of it. A chat app should be a room to think in — nothing between you and the model.
What you need stays: both frontier families in one place, conversations that branch when you want to explore, web search when you need citations, files when you need context. What you don't need is nowhere to be found. No newsletter traps, no AI “employees,” no dashboards you didn't ask for.
Tools, not theatrics.
· Features
Everything you need.
Nothing you don't.
Six features, one account. Each does its job, stays out of your way, and gets out of the page when you close the tab.
02
Branching conversations
Fork any reply to chase a different angle. Both threads persist — swap back when you want the original.
03
Web search, cited
Native grounding pulls live sources mid-answer. Every claim links back to where it came from.
04
File attachments
Drag in a PDF, a screenshot, a design file. The model reads it, you keep working.
05
Durable history
Every chat is searchable, exportable, yours. We don't train on any of it.
06
A free tier worth using
Three messages a day on every model, no card required. Pro when you're ready, not before.
· In motion
Same prompt.
Two answers, side by side.
Run GPT and Claude on the same input. Keep the one you like, branch the one you don't, or take a piece of each.
oran.chat / comparelive
You ·
GPT-4o
OpenAIClaude Opus 4.7
AnthropicStreaming in parallel…
Not a mockup — this is how the compare view behaves. Interaction lives in the app.
· Use cases
Three people.
One chat app.
oran.chat is built for thinking work. Here's what different kinds of thinking look like inside it.
persona
For writers
Draft in Claude's voice. Sharpen in GPT's. Branch to try a different tone without losing the original line.
- Multi-model drafting
- Branch to rewrite
- Attach a reference doc
persona
For strategists
Force two models to argue. Synthesize the disagreement. Use the places they agree as your shortest path.
- Side-by-side compare
- Web search for citations
- Export the thread
persona
For everyone else
One cleaner place to think out loud with GPT and Claude. No tab-switching, no premium paywall before you've even tried it.
- Three free messages / day
- Every model, every tier
- Keep what matters
· Pricing
Start free. Stay free if it suits you.
Two tiers. Same features, different generosity. Everything you actually need is in both columns.
Free
Enough to know whether you like it.
$0always
- 3 messages every day
- Every model, including Opus
- Branching, web search, file attachments
- Full conversation history
- No credit card, no dark patterns
Start free
Quota resets at midnight local.
Pro
For the days you'd rather not count.
$ —/ month
- Unlimited messages across all models
- Priority routing on busy hours
- Longer context windows
- Higher file upload limits
- Early access to new models
Go Pro
Cancel any time, keep your history.
· FAQ
A few fair questions.
What models do you support?
OpenAI's GPT-4o, GPT-4.1, and o4-mini, plus Anthropic's Claude Opus 4.7, Sonnet 4.6, and Haiku 4.5. New models show up within days of release.
Why not just use ChatGPT or Claude.ai directly?
Because then you can't compare. In oran.chat, one conversation can run on whichever model suits the turn — branch to another when you want a second read, swap mid-thread when the tone is wrong. One history. One search bar. No duplicate subscriptions.
How does the free tier work?
Three messages every day, on every model including Opus. No credit card, no lock-out after a trial. Quota resets at midnight local. If you hit the limit, the models stay visible — they just wait until tomorrow.
What's in Pro?
Unlimited messages, priority routing on busy hours, longer context windows, higher file-upload limits, and early access to new models as they ship. TODO: final feature copy + price.
Do you train on my conversations?
No. Conversations stay private to your account. OpenAI and Anthropic don't train on API traffic by default, and we don't log content for model training. You can export or delete anything, any time.
How does branching work?
At any reply, fork a new thread. The parent conversation stays exactly where it was — the branch gets its own context, its own model choice, its own future. Swap between branches like tabs; nothing is lost.
Can I export my conversations?
Yes. Every chat exports to Markdown from the settings panel. Bring your history anywhere; it's yours.
Still curious? hello@oran.chat
· A quiet invitation
Open the door.
Three messages on us, every day.
No credit card. No onboarding wizard. You're one link away from a chat with every model that matters.