Pricing & Wallet
Insighto.ai charges in two layers:
- A monthly plan with a generous allowance of every metered resource — queries, voice seconds, bots, and stored words.
- A pay-as-you-go Wallet that kicks in once any of those allowances is exhausted, billed per actual unit of usage.
You can also Bring Your Own Keys (BYOK) for LLMs and TTS — when you do, those line items drop out of the wallet equation because Insighto is calling your provider account directly.
This page covers the per-minute and per-query rates. For how to manage your wallet, see Billing & Wallet.
Monthly plan allowances
Each plan ships with a fixed bundle that resets on your billing-cycle anniversary:
| Allowance | What it counts |
|---|---|
| Queries | One text query (one user turn → assistant turn). |
| Voice seconds | One second of voice call duration. |
| Bots | Maximum number of assistants you can have configured. |
| Words | Total indexed words across all data sources. |
| Live agent seats | Live-agent inbox seats (add-on). |
| Concurrent calls | Max simultaneous voice calls. |
Plus binary feature flags: BYOK, Workflows, Calendar Pools, Custom voice, Forms, Custom tools, SIP, Human agents, Campaigns, Agency, White-label, Premium support.
Live status of your usage is on the Usage page.
Wallet (pay-as-you-go)
Once any allowance is exhausted, the Wallet takes over at the rates below. The Wallet is prepaid — top it up in any amount from $5 upward at Settings → Billing & Wallet. Quick tiles: $10, $25, $50, $100.
Voice — component breakdown
Every voice minute is the sum of four services. Base rates:
| Component | Base rate |
|---|---|
| Transcription (speech → text) | 1¢ / min |
| LLM (text → response) | 1¢ / min |
| Voice / TTS (response → audio) | 2¢ / min |
| Platform overhead | 2¢ / min |
| Total — baseline | 6¢ / min |
The baseline assumes a regular text LLM (gpt-4o-mini class, 1× multiplier) and Azure Voice (1× TTS multiplier). Switching the LLM or TTS provider multiplies the relevant line item.
TTS multipliers
| TTS provider | Voice multiplier |
|---|---|
| Azure Speech | 1× |
| Cartesia | 1.75× |
| ElevenLabs | 2.5× |
| PlayHT | 2.5× |
| Custom-cloned voice (ElevenLabs) | 4× |
LLM multipliers (voice)
| LLM family | LLM multiplier |
|---|---|
gpt-4o-mini, gpt-5.4-mini, gpt-3.5-*, deepseek-*, Llama 3.1 70B | 1× |
o3-mini | 2× |
gpt-4o (not -mini) | 10× |
Anything else (older gpt-4-*, Claude, etc.) | 20× |
Voice — pricing matrix
| Configuration | Multipliers (STT · LLM · TTS · Platform) | Cost breakdown (¢) | Total (¢ / min) |
|---|---|---|---|
| Regular LLM + Azure Voice | 1 · 1 · 1 · 1 | 1 + 1 + 2 + 2 | 6 |
| Regular LLM + Cartesia | 1 · 1 · 1.75 · 1 | 1 + 1 + 3.5 + 2 | 7.5 |
| Regular LLM + ElevenLabs | 1 · 1 · 2.5 · 1 | 1 + 1 + 5 + 2 | 9 |
| Regular LLM + PlayHT | 1 · 1 · 2.5 · 1 | 1 + 1 + 5 + 2 | 9 |
| Regular LLM + custom-cloned voice | 1 · 1 · 4 · 1 | 1 + 1 + 8 + 2 | 12 |
| Regular LLM + your TTS key (BYOK) | 1 · 1 · 0 · 1 | 1 + 1 + 0 + 2 | 4 |
o3-mini + Azure Voice | 1 · 2 · 1 · 1 | 1 + 2 + 2 + 2 | 7 |
o3-mini + ElevenLabs | 1 · 2 · 2.5 · 1 | 1 + 2 + 5 + 2 | 10 |
OpenAI Realtime (gpt-4o-mini-realtime-preview) | 0 · 10 · 0 · 1 | 0 + 10 + 0 + 2 | 12 |
OpenAI Realtime (gpt-4o-realtime-preview) | 0 · 46 · 0 · 1 | 0 + 46 + 0 + 2 | 48 |
Notes:
- BYOK voice is a hard zero. When your assistant's TTS voice comes from your own provider connection, the voice line item is multiplied by 0. The LLM line is not auto-zeroed by BYOK — that depends on whether the platform-managed OpenAI key or your own OpenAI key ran the turn.
- Realtime models bundle STT + TTS into the LLM cost. When the assistant type is Realtime OpenAI, the platform doesn't run separate transcription or TTS — the LLM line absorbs all three.
- Custom-cloned voices cost 4× the base TTS rate.
Chat — per-query rates
Each text query (one user turn → assistant turn) is billed at 1.5¢ × llm_multiplier:
| LLM model | Multiplier | Cost per query |
|---|---|---|
gpt-4o-mini, gpt-5.4-mini, gpt-3.5-*, deepseek-* | 1× | 1.5¢ |
o3-mini | 2× | 3.0¢ |
gpt-4o (not -mini) | 10× | 15¢ |
Anything else (gpt-4, gpt-4-turbo, Claude, etc.) | 20× | 30¢ |
Pick the LLM in Build → Assistants → [your assistant] → Model. gpt-4o-mini is usually the sweet spot — strong reasoning at 1.5¢ per query.
Other usage
| Resource | Rate |
|---|---|
| Stored words (data sources) | Included up to plan limit; overage charged per 100k words / month |
| Outbound SMS | $0.015 / message (US baseline); varies by destination |
| Outbound voice (campaigns) | Same per-minute rates as inbound + telephony toll |
| Telephony (Twilio / Plivo / Telnyx) | Pass-through at provider rates |
Example calculations
A 5-minute voice call with regular LLM + Azure Voice:
5 min × 6¢ = 30¢ deducted from wallet
A 10-minute voice call with o3-mini + ElevenLabs:
10 min × 10¢ = $1.00 deducted from wallet
100 chat queries on gpt-4o:
100 × 15¢ = $15.00 deducted from wallet
A 5-minute Realtime call (4o):
5 min × 48¢ = $2.40 deducted from wallet
Every per-call deduction is itemised on the conversation record — the billing details JSON stores the exact component costs and multipliers used. You can read these back via the conversation API.
BYOK — Bring Your Own Keys
If you've configured BYOK Credentials (Settings → BYOK Credentials), Insighto calls your provider account for the supported providers. The conversation is marked billed_to = byok and neither plan credits nor wallet are touched at all.
BYOK key slots:
- OpenAI — used for LLM (chat completions, embeddings) and the Realtime API.
- ElevenLabs — used for TTS when an ElevenLabs voice is selected.
- Azure Speech — used for STT and TTS when an Azure voice / STT is selected.
BYOK does not cover:
- Deepgram STT, Cartesia TTS, PlayHT TTS, Anthropic — no BYOK slot today.
- Telephony (Twilio / Plivo / Telnyx) — always carrier-billed.
- Platform overhead (2¢/min on voice) — kept by us for orchestration.
- Add-on features (Workflows, Calendar Pools, etc.) — gated by plan tier, not wallet.
When BYOK is fully on (master toggle on, all three slots populated) and the assistant uses only providers covered by your keys, the runtime sets deduct_credits = false — the wallet is untouched, the plan credits are untouched, and the conversation is filed as billed_to = byok.
→ Set it up: BYOK Credentials
Auto-recharge
The wallet has auto-recharge fields (threshold, amount, enabled) wired into the schema, but no UI drives them for direct customers today. Top up manually for now.
Agency rebilling
If you're an agency, you set your own plan + wallet pricing for your subaccounts, with optional cost multipliers on top of Insighto's base rates. See Rebilling.
Where to next
- Billing & Wallet — how the plan + wallet flow works day to day.
- BYOK Credentials — drop LLM and TTS costs by routing through your own keys.
- Voice settings — pick the STT and TTS providers that drive the voice line.