FAQ

How usage limits work

What draws on your plan’s allocation, and what happens when you reach a limit.

Not every prompt asks the same of your AI models. The model you choose, how long your conversation runs, and whether you’ve uploaded documents all affect how much of your plan’s monthly allocation a session uses.

What affects your allocation

Three things determine how much of your allocation a conversation uses.

The model you choose. Flagship models are the most capable and use more of your allocation per exchange. Fast models — Claude Haiku and GPT-5.4 mini — use significantly less. For routine work like drafting, summarizing, or quick research questions, fast models often deliver what you need at a fraction of the cost to your allocation. Save flagship models for the analysis that requires them.

How long your conversation is. Every time you send a message, the model reads your entire conversation history as context. A thread with forty exchanges costs more per message than one with four, because the model is processing everything that came before. If you’re working through a complex research question, starting a new conversation for each distinct line of inquiry is a good habit.

Documents you’ve uploaded. Document content is included in the model’s input the same way conversation history is. Longer documents use more of your allocation per exchange than shorter ones.

How limits work

Your plan includes a monthly usage allocation. Bells Up AI also tracks usage on a daily basis so that one heavy session doesn’t consume a disproportionate share of your team’s monthly allocation.

Most teams won’t come close to their limits. If yours does, limits work in two stages rather than cutting you off entirely.

Stage one: fast models only. When your team’s daily usage reaches its threshold, flagship and mid-tier models become temporarily unavailable. You can still send messages — using Claude Haiku, GPT-5.4 mini, and other fast models. A banner appears at the top of the interface to let you know. Daily limits reset each day.

Stage two: access paused. If usage on fast models continues beyond the daily fast-model threshold, AI access pauses until the daily reset. This is rare in practice.

The same two-stage logic applies to monthly limits independently of daily limits. Whichever limit — daily or monthly — is reached first determines your state.

Teams and shared allocation

Usage is tracked at the organization level, not per person. Everyone on your team draws from the same pool. A team with more seats has a proportionally larger allocation — adding seats scales your team’s monthly budget accordingly.

Organization administrators can set individual usage caps for specific team members if needed.

How to know where you stand

When your team is in normal mode — under all limits — nothing changes and no indicators appear. You won’t see usage information unless it’s relevant.

When you enter restricted mode (fast models only), a banner appears at the top of the chat interface describing your current state and when it resets. Models that are temporarily unavailable show a lock icon in the model selector.

Note

Panel of Experts, Complaint Analysis, and other workflow tools run multiple model calls in parallel. These tools use more of your allocation per session than a single conversation exchange. They are designed for the analysis that requires them — use them when the task warrants it.

Summary

State What it means What you can do
Normal Under all limits Full access to all models on your plan
Restricted Daily or monthly threshold reached Send messages using fast models
Paused Fast-model threshold also reached Wait for the daily reset