Credits and Cost

What an Alphio AI action actually costs in Credits, and what to do when costs surprise you.

---

How much does a chat query cost in Credits? A single chat query is one action class — a single Alphio AI reply consumes a Credit amount that depends on what the reply did:

• A plain text answer with no tool calls is the cheapest case.
• An answer that invokes one or more tool calls (data lookup, screener, chart fetch, on-chain query) costs more — each tool call adds to the action cost.
• An answer that streams long context or chains multiple internal reasoning steps costs more than a one-shot answer.

How much does a Tasks job cost? Tasks are asynchronous AI jobs and cost more than a one-shot chat reply because they run for longer and call more tools:

• Light Tasks (e.g. simple monitoring) — a small multiple of a chat query.
• Medium Tasks (scheduled analysis, multi-source research) — substantially more, scaled to depth.
• Heavy Tasks (long multi-source research runs) — the most expensive class.

Long-running Tasks may consume Credits incrementally as they progress. Your actual consumption is always visible in your Credits activity (Billing & Usage on the web, or the Usage page on mobile web and in the app).

A research run seems expensive — why? Cost tracks the work performed, and a long research Task does far more than a chat reply:

• Many tool calls — multiple data sources are queried.
• Long context — the model reasons over a much larger working set than a short chat.
• Multiple internal reasoning passes — the agent plans, executes, and refines.

If you only need a quick answer, use a regular chat query instead — it's much cheaper. Reserve Tasks for when you actually want the multi-source, long-form output.

Where can I see what an action cost? Check your Credits activity — Credits activity on the Billing & Usage page (/account/billing), or My Credits on the Usage page — which lists this month's consumption by source (Chat, Trade Agent, backtests, and so on).

I'm running out of Credits — how do I extend? Three paths, in order of preference:

• Custom Credits — buy a one-time Credits top-up on the Top-Up Credits page (requires a paid plan; not available during a trial). Recharge Credits roll over between billing cycles and are valid for 1 year from purchase.
• Upgrade tier — moving up a tier gives more Monthly Credits each cycle.
• Wait for cycle refresh — Monthly Credits refresh on your billing cycle's renewal date. (The Free tier's 400 Credits are a one-time allowance and don't refresh.)

Limited-time Credits (from promos and rewards) can also extend your runway but expire on a fixed date; the Usage page lists each bucket's expiry date.

Credits used but no useful answer — refund? There is no self-serve refund for an answer you found unhelpful, and no way to request one from the Billing & Usage or Usage pages. An answer that completed is treated as a delivered action — Credits track the work the model performed, not whether the output satisfied you.

What does protect you is that failed calls aren't charged at all:

• If the model call errors, Credits are never deducted — the deduction only runs after a successful reply.
• If a stream dies mid-answer, you are not billed, even though the work was partly done.
• If the model returns no token usage, the system refuses to bill rather than estimating a cost.

So a genuine failure costs you nothing. A completed-but-disappointing answer is a successful call, and it is billed.

Also note Custom Credits are non-refundable after purchase — see Custom Credits.

If you believe you were charged in error (e.g. an action billed that never produced a reply), contact contact@alphio.ai with the timestamp and the Credits activity entry (Billing & Usage, or the Usage page). Support can look into it, but a goodwill adjustment is discretionary — it isn't something you're entitled to, and unhelpful output on its own isn't grounds for one. To keep costs down, prefer a regular chat query over a Task when you only need a quick answer.