Browse the documentation

Documentation

AI modes

Updated 3 min read

Ask AI about this page

Every account runs on one of three AI modes: Economy, Power or Max. The mode sets which class of models transcribes and analyzes your calls, and the per-call analysis fee, which covers both the transcription and the AI analysis. Per-minute rates are the same on every mode. Change it on the Dashboard, under Billing or by API.

What a mode is#

Analysis has a per-call cost, and different traffic deserves different depth. A mode bundles two choices into one setting the account controls: which class of models transcribes and analyzes a call, and what that analysis costs. Per-minute call pricing is identical on every mode. The mode changes only the per-call analysis fee, and that fee covers both the transcription and the AI analysis of the call.

The three tiers. Live fees are on pricing.
ModeWhat it doesAnalysis fee
Economy The wide net: fast, lean models on every call. Built for sampling everything. $0.015/call
Power (default) The daily driver: balanced transcription and analysis for production traffic. $0.02/call
Max The deep read: our most capable models on every single call. $0.05/call

Where the selector lives#

Two places in the portal, backed by the same setting. The Dashboard carries a slim strip under the stat tiles with the three tiers as pills and their fees. One click switches the tier. The Billing page carries the full card with each tier's description. Only the account's users can change it, and every switch is recorded in the audit log.

The AI mode card on the Billing page: Economy (the wide net, $0.02 analysis per call), Power (the daily driver, $0.05, marked current) and Max (the deep read, $0.15), with a note that per-minute pricing is identical on every tier.
The AI mode card on Billing: the three tiers and their fees. One click switches.

How switching behaves#

  • A switch applies to calls analyzed from that moment, with no restart or delay.
  • A call already in flight keeps the tier it started with, and is billed under it.
  • The tier is stamped on the call when it is processed, so the fee always follows what ran, never the setting at some later time.

How it shows on the CDR and the ledger#

Three places name the tier, so a fee can always be reconciled against what produced it: the AI mode column on the Calls table, the tier chip on the call detail's Processing panel, and the ledger line for the analysis fee, for example Call analysis (economy). Calls from before the feature carry no tier and keep the plain description.

By API#

Each portal action is one API call. The tiers and fees are also public, like /plans.

read, then switch

# the tiers, their fees, and which one is current
curl -s https://ovs.telecomsxchange.com/api/customer/ai-modes \
  -H "Authorization: Bearer $TOKEN"

# switch the account to Economy, applies to new calls instantly
curl -s -X PUT https://ovs.telecomsxchange.com/api/customer/ai-mode \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{"mode": "economy"}'

A common pattern: run Economy to spot-check a wide share of traffic cheaply, and raise the tier on the routes or periods that deserve the deep read. The API reference carries the full request and response shapes, Concepts explains every field the analysis returns whichever tier produced it, and the current fees sit beside the plans on the pricing page.

Frequently asked questions#

What is the difference between Economy, Power and Max?

Each tier maps to a class of transcription and analysis models and a per-call analysis fee. Economy is the wide net: fast, lean models at the lowest fee, built for sampling everything. Power is the balanced default. Max runs the most capable models on every call, for routes where missing one call costs more than checking them all. Per-minute rates are identical on every tier.

When does an AI mode change take effect?

Instantly, for calls analyzed from that moment. A call already in flight keeps the tier it started with, and its fee follows that recorded tier, so switching mid-day can never mis-bill a running call.

How do I see which mode a call was billed under?

Three places: the AI mode column on the Calls table, the tier chip on the call detail's Processing panel, and the ledger line for the analysis fee, which names the tier, for example "Call analysis (economy)". Calls from before the feature show no tier and keep the flat fee.