The smartest output wins.
Three top AIs draft. You pick the synthesizer. Get the best of Claude, GPT, and Gemini — combined into one output.
Built for every kind of thinking.
One quick question. A coding marathon. A whole agent workspace. Same platform, four ways in.
Multi AI
Three models draft, debate, and synthesise — you pick the synthesizer.
Single
One model, one output. Fast.
Cowork
Specialist roles — Reasoning plans, Coding builds, Design refines.
Advanced Thinking
A deliberation layer on top of any mode — three models deliberate, an arbiter decides. All plans.
Introducing Cowork
A workspace
for your models.
Beyond chat. Cowork lets your models specialise into roles, read your files, and pass work to each other — without you copy-pasting between tabs.
What's inside.
Three specialist roles
Reasoning plans, Coding builds, Design refines — each role runs the model you pick.
Sequential orchestration
Reasoning gets the first and last say. Coding and Design run only when the task signals require them.
Workflow handoff
Every step is visible in the thread. Skipped roles render as a small pill so the routing decision is transparent.
Simple pricing.
Free
- Rolling 30-day allowance
- Standard models + Advanced Thinking
- Single + Multi AI Comparison
Plus
Recommended- Frontier models
- Multi AI Debate + Cowork
- + everything in Plus →
On the roadmap.
Latest shipped, then what's next, then where we're going.
Now available
Advanced Thinking
A toggle below the composer. Three models deliberate, critique each other, an arbiter resolves the definitive output. Available on every plan — the arbiter scales with your tier.
Also live
Variable usage top-ups
Pay between £1 and £100 of usage in one click. A platform fee is included in each top-up — higher amounts offer better value. The exact fee is shown on every purchase before you commit. Top-ups are valid for 30 days.
Next
Pro tier
Higher monthly allowance, Opus 4.7 as the AT arbiter, team workspaces, API access.
Future
AISynvex Forge
Tier-split autonomous agent. Free and Plus get a single-agent flavour: one model takes your task, executes inside a secure sandbox, and delivers the result. Pro gets the full three-model debate — three models deliberate on every decision, then execute and review together.