# Concurrent agents

Concurrent agents: How many agent sessions can be running for your workspace at the same time, the actual unit CodeHerder's plans are metered on.

Source: https://codeherder.com/glossary/concurrent-agents/

Glossary · The money

How many agent sessions can be running for your workspace at the same time, the actual unit CodeHerder's plans are metered on.

## What it is

Concurrent agents is a count of how many sessions your workspace can have actively running at once. It has nothing to do with how many tasks you file in a month or how many people are on your team. A plan's limit caps that number, and it's the dial that decides how much work can genuinely happen in parallel. Two teams filing the same number of tasks a month can need very different concurrency limits, depending on whether that work arrives steadily or all in a rush.

## Why it matters to you

This is the number that actually determines throughput. Headcount doesn't enter into it. A five-person team capped at two concurrent agents can still only run two tasks in parallel at a time; a solo founder with a higher concurrency limit can have several tasks moving at once. Size your plan around how much you want happening simultaneously. How many people will ever log in doesn't matter, since adding people to a workspace costs nothing by itself.

## How it works in CodeHerder

CodeHerder's plans are metered on concurrent agents rather than seats, and you separately bring your own model provider key, so you're never billed for tokens through the platform itself. What each plan tier includes, and what happens when a task arrives with every slot already busy, is on the plans and limits doc, and the pricing page has the actual tiers and their limits side by side.

- [Pricing →](https://codeherder.com/pricing/)
- [Plans and limits →](https://codeherder.com/docs/plans-and-limits/)

## What people get wrong

Concurrency and total volume sound like the same thing, and they're not: a low concurrency limit doesn't mean fewer tasks get done over a month, it means fewer of them run at the exact same moment. A team that mostly files work throughout the day, rather than all at once, can get by on a lower limit than the raw task count might suggest, since the queue simply moves through tasks one after another instead of stalling. The number worth watching is how many tasks pile up waiting for a slot. The total you filed matters far less.

Keep going

## Related terms

[Agent session One running instance of a harness working a task on a device, the actual process you can watch in real time.](https://codeherder.com/glossary/agent-session/) [Context window How much text a model can actually hold in mind at once, and one of the biggest levers on what a task costs to run.](https://codeherder.com/glossary/context-window/) [Device A machine you've registered, your laptop, a spare workstation, or a cloud instance, that CodeHerder can actually dispatch task work to.](https://codeherder.com/glossary/device/)

[← See every term](https://codeherder.com/glossary)