How RelayMRL Works
Four AI models, one conversation. Ask a question and watch them think together.
What is RelayMRL?
RelayMRL puts four leading AI models — Claude, ChatGPT, Perplexity, and Grok — into the same conversation. They can see each other's answers and naturally agree, disagree, or build on the reasoning.
Think of it like having four smart advisors at a table instead of one. Where they agree, you know the signal is strong. Where they disagree, you've found something worth digging into. Either way, you get a more complete picture than any single model can give you.
No copy-pasting between tabs. No losing context. Just ask your question and let them work.
Getting started
- 01
Pick your models
All four models are active by default. Want just two? Toggle models on or off from the settings in the composer. You need at least two active to keep the conversation multi-perspective.
- 02
Ask your question
Type anything and hit Ask All to send it to every active model at once, or click a specific model name to ask just one. You can attach images, PDFs, or even connect a GitHub repo for the models to reference.
- 03
Relay between models
This is where the magic happens. After a model responds, use the relay bar to pass that answer to another model. They'll react to it — agreeing, pushing back, or adding something the first model missed. You can keep relaying to build a full picture.
- 04
Get the bottom line
Once the models have weighed in, RelayMRL can synthesize everything into a clear recommendation — where they agree, where they don't, and what you should actually do.
The models
Careful, critical thinker. Strong at analysis, catching edge cases, and nuanced reasoning. Will tell you what others might be glossing over.
Practical and broad. Great at actionable plans, structured thinking, and pulling together the big picture from messy inputs.
Research-first with real-time web access. Brings data, citations, and current information that the other models don't have.
Direct and opinionated. Good at cutting through noise, challenging assumptions, and giving you the take nobody else will.
Each plan uses different model versions. Free uses fast, efficient models. Pro upgrades to Claude Sonnet, GPT-5, Sonar Pro, and Grok 4. Ultra gets the most powerful versions available — Opus, GPT-5.5, Sonar Reasoning, and Grok 4.3.
Features
Ask All
Send your question to every active model at once. Responses arrive in parallel so you can compare them instantly.
Relay
Pass one model's answer to another for a targeted reaction. This is how real disagreements and insights surface.
The Bottom Line
A synthesized recommendation that weighs all perspectives — where they agree, disagree, and what the best move is.
Response length
Control how verbose the models are. Normal, Short, or Brief — set it before your first message or change it anytime.
Image & PDF uploads
Attach screenshots, photos, diagrams, or documents. The models analyze what they see and reference it in their answers.
GitHub integration
Connect your GitHub to load repository context directly into conversations. Models can reference your actual code, not hallucinated versions of it.
Memory
RelayMRL remembers context across conversations. It picks up relevant details from past threads so you don't have to repeat yourself.
Generate
Turn conversations into action plans, checklists, email drafts, design briefs, or research plans — generated from the full context of the discussion.
Threads
Every conversation saves automatically. Pick up where you left off, or start fresh whenever you want.
Deep Research mode
Switch to Deep Research when thoroughness matters more than speed. Models go deeper and the platform generates analysis briefs.
Tips
Use relay strategically. If Claude gives a strong technical answer, relay it to Grok — they'll challenge the parts Claude was too polite to question. If Perplexity brings data, relay it to ChatGPT to build an action plan from it.
Don't always Ask All. Sometimes you want one model's deep take before bringing in the others. Start with the model best suited to your question, then relay.
Use Brief mode for quick checks. Not every question needs a four-paragraph response. Set the response length to Brief for fast, punchy answers.
Attach context. The more the models know, the better they respond. Upload a screenshot of what you're working on, connect a GitHub repo, or just paste relevant details into your message.
Your data
- ✓ Conversations sync securely to your account — accessible from any device
- ✓ We don't train any models on your data
- ✓ Uploaded files are processed for the conversation and not stored permanently
- ✓ Delete any thread anytime — it's gone for good
What's new
Recent updates to RelayMRL.
May 2026
- Grok joins the table. Four models now — Claude, ChatGPT, Perplexity, and Grok. Toggle any combination from the composer.
- Response length controls. Set responses to Normal, Short, or Brief right from the start — no need to begin a thread first.
- Ultra tier. Opus 4.8, GPT-5.5, Sonar Reasoning Pro, and Grok 4.3 for users who want the most capable models available.
- Memory. RelayMRL now remembers relevant context across your conversations.
- GitHub integration. Connect your account and load repos directly into conversations — including private ones.
March 2026
- Thread sync. Conversations now sync to your account instead of just living in your browser.
- Generate actions. Turn any conversation into action plans, checklists, email drafts, design briefs, or research plans.
- Deep Research mode. A dedicated mode for complex questions where thoroughness matters.
- Image and PDF uploads. Attach files directly — all models can analyze them.
February 2026
- Launch. RelayMRL goes live with Claude, ChatGPT, and Perplexity. Ask All, Relay, and The Bottom Line.
Ready to try it?
Get started for free