Vannus Room launches: one workspace, multiple AI models, hard compute caps

Today we’re opening Vannus Room — a multi-model AI workspace built on two principles you don’t see together very often: choice of model, and a hard cap on what you spend.

The first principle is obvious. Different models are better at different tasks. The second is the one that takes a little explaining. Every other subscription AI product we’ve evaluated either gives you unlimited use at a high flat price (which means the heavy users subsidize the light users) or charges per-token and lets your bill float (which means a bad day or a runaway prompt can ruin your month). We don’t do either.

What Room actually does

Room is a chat workspace where every conversation is metered against a fixed monthly compute budget. The budget is set by your tier. The number of prompts you can send depends on how heavy each conversation is. A short factual question costs almost nothing. A multi-thousand-token research thread costs more. Either way, the cap is the same number every month.

When you reach your cap, Room pauses for you until the next billing cycle. We never bill you over your tier. We never roll usage over. We never charge per-token surprise fees.

The free Vannus platform stays free. Catalog, methodology, ROI calculator, RFP builder — none of those require an account, and that won’t change. Room is a paid product because the compute behind it isn’t free, and we’d rather price honestly than offer a free tier that nudges you into upgrade pressure.

Why caps matter more than they look

The single most underrated risk in AI tooling in 2026 is unpredictable monthly cost. The standard pricing model in this category is “pay-as-you-go.” That sounds reasonable until you watch a team accidentally burn $2,000 in three days because someone left a script running, or until your monthly LLM bill is the kind of line item your CFO won’t stop asking about. We’ve seen both happen.

A hard cap removes that risk entirely. You know what you’ll pay. If you need more, you upgrade or add a seat. If you don’t, your cost is fixed. That’s it.

What’s coming next: multi-model collaboration

Room launches today with single-model conversations — pick the model, ask the question, get the answer. That’s table stakes. The bigger bet is on the next layer, which we’re building now and expect to ship in the coming months.

The premise: for hard questions, the best answer doesn’t come from one model. It comes from multiple models working on the same task in different ways — one drafting, another critiquing, a third double-checking. Vannus’s neutrality-driven routing engine is positioned to do this in a way that single-vendor products can’t, because we’re not optimizing for any one provider’s revenue.

We’ll share more on this when it’s closer to launch. For now, every Pro+ subscriber gets it automatically when it ships, no upgrade required.

The thing we’re not doing

One pattern we’re deliberately avoiding: the “use our credits to access cheap models, then upgrade for the expensive ones” trap. Pro and Pro+ include the same neutrality engine, the same workspace persistence, the same compute-cap protection. The difference is which models you can route to and how much compute budget you have. That’s a difference of volume, not quality of treatment. The free platform and the paid tiers all run on the same trust principles. We don’t cripple cheap tiers to push you up.

The other pattern we’re avoiding: data harvesting. Chat content in Room is never used to train Vannus models. The third-party LLM providers we route to commit, in their API terms, not to train on customer inputs either. We don’t guarantee third-party compliance, but we choose providers whose API terms make the right commitment. The Privacy Policy spells out exactly how this works.

If you’re ready

If you’d rather start with the free Vannus surfaces — the Tools catalog, the elimination-first methodology, the Build My Stack walkthrough — those stay open with no signup required, as always.

And if you have a larger stack rationalization problem that’s outside what Room is designed for, the Concierge audit is available at $7,500 for the full founder-led engagement.