How AI tokens are spent
There is one balance, not one per module. That is the whole design, and it explains both the good and the annoying parts.
Every AI feature in Lodos draws from the same wallet. Ask Data Analyst a question, render a scene in Yeşilçam Studios, run a site audit in Web Analyzer, let Sam answer a support question — all of it comes out of one monthly token balance attached to your account.
That is unusual. Most suites sell you credits per product, which means you end up with 400 unused video credits and no analysis credits on the day you need analysis. One wallet avoids that. It also means a careless afternoon in one module can eat the month.
What each plan includes
Tokens refill on the first of every billing month and do not roll over. The allocation is per user, not per workspace.
- Standard (free): 250,000 tokens a month, 1 GB storage, all thirty modules, single user.
- Basic ($30/user/month): 5,000,000 tokens, 30 GB, team invites.
- Pro ($99/user/month): 15,000,000 tokens, 50 GB, uncapped workspaces and automations.
250,000 sounds small next to 5 million, and it is — but it is enough to ask Data Analyst a few dozen real questions in a month. It is not enough to render video. That gap is deliberate: the free plan is meant to let you evaluate everything, not to run a studio.
The part nobody notices: the multiplier
The same operation does not cost the same on every plan. Each module operation has a base cost, and your plan applies a multiplier to it before the deduction:
- Basic pays the base cost — multiplier 1.0
- Pro pays 20% less — multiplier 0.8
- Standard (free) pays 30% more — multiplier 1.3
So a job that costs 10,000 tokens on Basic costs 8,000 on Pro and 13,000 on the free plan. If you are close to the line between Basic and Pro and you run AI-heavy work, the effective token difference is larger than the sticker prices suggest: Pro gives you three times the allocation and each operation costs a fifth less on top of that.
Running out mid-month
When the balance hits zero the AI operation stops and you get a message saying so — nothing is charged automatically and no module breaks. Everything non-AI keeps working exactly as before. Boards, files, meetings, time tracking, the browser: none of them touch the wallet.
To top up, buy a pack. These are one-time purchases, not a subscription change, and the tokens sit on top of your monthly allocation:
- Starter — 1,000,000 tokens, $3. Available on every plan, including free (at $6 there).
- Value — 5,000,000 tokens plus a 10% bonus, so 5,500,000 for $10. Paid plans only.
- Power — 15,000,000 plus 20%, so 18,000,000 for $25. Paid plans only.
The bonus makes the larger packs meaningfully cheaper per token, and the Power pack is roughly half the price per token of the Starter pack. If you know you will use them, buying one Power beats buying six Starters over a quarter.
Finding out where the month went
The wallet keeps a per-operation ledger, and it is the first place to look when a month disappears faster than expected. Nine times in ten it is one of three things:
Video and image generation. Yeşilçam Studios and Fashion TryOn are orders of magnitude more expensive per call than text. A single video render can cost more than a week of database questions. This is not a Lodos pricing decision — it is what the underlying models cost — but it does mean these two modules deserve their own mental budget.
An automation in a loop. Automation flows can call AI steps, and a flow triggered by "any record changes" that itself changes a record is a loop that bills. Automation shows a run count per flow; a flow with four figures next to it after two days is the answer.
A large document parsed repeatedly. Asking three follow-up questions about the same 200-page PDF costs three times, because each question re-reads the context. Ask the compound question once instead of drip-feeding it.
Spending less without doing less
Two habits cut usage noticeably, and neither costs you anything in quality.
Be specific in the first prompt. "Summarise Q3 revenue by region and flag anything more than 15% below plan" costs one operation. Asking for the summary, then the regions, then the outliers costs three, and the third answer is usually worse because the context has drifted.
Narrow the data before you ask. Data Analyst answers questions about whatever you point it at. Pointing it at one table instead of the whole schema produces a cheaper, faster and more accurate answer, in that order of magnitude.
One thing that does not help: turning AI modules off in the sidebar. Hiding a module changes what you see, not what you are billed for. Tokens are only deducted when an operation actually runs.
Questions people actually ask
Do unused tokens roll over to next month?
No. The allocation resets at the start of each billing month. Purchased packs behave differently — those stay on your balance until you spend them.
Is the token balance shared across a workspace?
No, it is per user. Five people on Basic have five separate 5,000,000-token balances, and one person running heavy jobs does not drain anyone else.
What happens when I run out mid-task?
The AI operation stops and tells you why. Nothing is charged automatically, no work is lost, and every non-AI module keeps working normally until you top up or the month resets.
Why did the same operation cost less this month?
Most likely your plan changed. Pro applies a 0.8 multiplier to every operation and the free plan applies 1.3, so identical work costs 20% less on Pro than on Basic and 30% more on Standard.
