Device License Pricing
Full-stack AI voice solution for hardware, licensed per device per year. Includes ASR + LLM + TTS + Visual Understanding — all in one SDK.
Choose Your Tier
Each device is bound to a single License, valid for 1 year from activation. Voice + LLM inference + Visual understanding all included, integrated with a single SDK.
- 9 hours voice time
- 2,700,000 LLM inference tokens
- 50 rounds visual conversation *
- ASR + LLM + TTS + Vision
- Single SDK integration
- 1 purchase per device per year
- 19 hours voice time
- 5,700,000 LLM inference tokens
- 200 rounds visual conversation *
- ASR + LLM + TTS + Vision
- Single SDK integration
- 1 purchase per device per year
- 60 hours voice time
- 18,000,000 LLM inference tokens
- 1,000 rounds visual conversation *
- ASR + LLM + TTS + Vision
- Single SDK integration
- 1 purchase per device per year
Basic Tier vs OpenAI Realtime: Same 9h Conversation, ~770× Cheaper
| Voice Time | LLM Tokens | Visual Conv. | Price | |
|---|---|---|---|---|
| Adventists AI | 9 hours | 2.7M | 50 rounds | $0.99 |
| OpenAI gpt-realtime (end-to-end audio) | 9 hours | included in audio tokens | — | ~$762 |
A 9-hour voice conversation on OpenAI's gpt-realtime API costs ~$762 (audio in/out billed by token) — Adventists AI License delivers the same usage for $0.99, plus visual conversation. All three tiers save ~93% vs. buying the equivalent à la carte (see the "à la carte" tag on each card).
* OpenAI gpt-realtime estimated at $32/M audio input + $64/M audio output, with 9 hours of bidirectional conversation modelled as audio tokens at OpenAI's published Realtime token rates. See the full breakdown in the industry comparison below.
10 free Basic Licenses included upon signup (valid for 1 year). All devices under one product must use the same tier.
* Visual conversation rounds only count visual understanding API calls. Voice time and LLM tokens consumed during the conversation are deducted from their respective quotas normally. Based on 720p (1280×720) standard resolution; higher resolutions are scaled proportionally by pixel area.
Pay-As-You-Go, Only for What You Use
When License quotas are exhausted, purchase resource packs to continue. Shared across all devices in a product, valid until fully consumed.
| Resource | Billing Unit | Unit Price |
|---|---|---|
| LLM Inference | 1 Billion tokens (i.e. $0.28 / million tokens) | $280 |
| Voice Time (ASR + TTS) | 100 hours | $133 |
| Visual Conversation * | 1,000 rounds | $20 |
| Voice Cloning | voice / year | $150 |
* Visual conversation rounds only count visual understanding API calls, excluding voice and LLM consumption during the conversation. Based on 720p; 1080p ≈ 2.25 rounds, 4K ≈ 9 rounds. Actual usage may vary per API response.
LLM Flagship Model Unit Price (USD / million tokens)
* Based on publicly listed API prices as of May 2026. Our $0.28 is a uniform input/output rate; every other vendor is normalized as (input + output) / 2. CNY prices converted at $1 ≈ ¥7.2 where applicable.
Full-Stack Cost at Equivalent Usage
Based on Basic tier usage (9 hours voice + 2.7M tokens), comparing the equivalent full-stack cost using each vendor's 2026 flagship LLM. All non-Adventists LLM prices are normalized as (input + output) / 2.
* Based on publicly listed API prices as of May 2026. Each vendor's LLM uses the latest flagship version, normalized as (input + output) / 2. Our $0.99 is the Basic License price (uniform $0.28/M input/output). OpenAI bar is truncated. See the table below for detailed breakdowns.
| Vendor / Flagship LLM | Billing Model | ASR (9h) | LLM (2.7M, avg in/out) | TTS (9h) | Total |
|---|---|---|---|---|---|
| Adventists AI (in-house) | Full-Stack License | — | — | — | $0.99 |
| DeepSeek V4 Pro Stack | 3rd-party ASR/TTS + LLM | $0.37 | $1.78 ($0.66/M) | $1.00 | $3.15 |
| Doubao Seed 2.0 Pro | Volcano Engine full-stack | $0.25 (¥1/h × 1.8h) | $3.59 ($1.33/M) | $2.70 (¥3/10K-chars × 64.8K chars) | $6.54 |
| Zhipu GLM-5.1 Stack | 3rd-party ASR/TTS + LLM | $0.37 | $5.78 ($2.14/M) | $1.00 | $7.15 |
| Alibaba Qwen3-Max | Bailian full-stack | $0.37 (Paraformer) | $6.32 ($2.34/M) | $1.00 (CosyVoice-flash) | $7.69 |
| Kimi K2.6 Stack | 3rd-party ASR/TTS + LLM | $0.37 | $6.70 ($2.48/M) | $1.00 | $8.07 |
| Claude Sonnet 4.6 Stack | 3rd-party ASR/TTS + LLM | $0.37 | $24.30 ($9.00/M) | $1.00 | $25.67 |
| OpenAI gpt-realtime | Realtime API per token | Audio input $32 + output $64 / M tokens (avg $48/M) | ~$762 | ||
* Doubao Seed 2.0 Pro: Volcano Engine 0-32K input tier (¥3.2 in + ¥16 out, blended ¥9.6 ≈ $1.33/M); ASR/TTS priced with Volcano Engine's own rate card — 9 hours of dialogue is modelled as a 1:4 split typical of AI-companion workloads (1.8h user-side ASR + 7.2h AI-side TTS). ASR: 1.8h × ¥1/h (Doubao Streaming ASR 2.0); TTS: at a natural Mandarin pace of 150 chars/min, 7.2h ≈ 64,800 chars × ¥3/10K chars (Doubao Speech Synthesis 2.0). Alibaba Qwen3-Max official rate $0.78/$3.90 (avg $2.34/M), paired with Paraformer + CosyVoice-flash. DeepSeek / Kimi / Zhipu / Claude do not offer native ASR/TTS; estimated using Alibaba Paraformer ($0.37/9h) + CosyVoice-flash ($1.00/9h). OpenAI uses gpt-realtime standard audio tokens. CNY conversions at $1 ≈ ¥7.2.
Ready to Get Started?
Whether you're integrating APIs or building hardware, we're ready.