Planned cost
$611
For selected filtersLoading planCost dashboard
Editable monthly model for feature-level AI, speech, and audio delivery costs.
Loading plan
Segment filters
Mocked event-time filters. User type is counted at the moment of usage, so trial usage and later paid-yearly usage stay separate.
Device platform
All devices
App version
All versions
Device country
All countries
User type
All user types
Last 30 daysAll devicesAll versionsAll countriesAll user types100% usage slice
Actual cost
$733
For selected filtersLoading actual usageCost delta
+$122
+20% vs planActual minus plannedMost over-plan feature
Roleplay / Conversation Practice
+$11814 costing rowsMonthly cost shape
Visual from actual row totals. The tables are the source.
Guided Lesson - Vocab
Word card -> learner speaks -> pronunciation score -> result shown
Actual vs planned$41.18-$3.83
| Costing sub-feature | Pricing and source | Planned usage | Actual usage | Planned cost | Actual cost | Delta | Note |
|---|---|---|---|---|---|---|---|
| Vocabulary pronunciation scoringGoogle Cloud Speech-to-Text V2, chirp_3 Learner audio -> speech scoring -> result. Cost = sessions x attempts x seconds. | $0.016 / processed minute, first 500k minutesGoogle STT | 18,500 sessions x 2.4 attempts x 3.8 sec / 60 x $0.016 = $44.99 | Loading actual usage Monthly sessions16,260sessions Attempts / session2.302attempts Seconds / attempt4.124sec Rate / min0.016$ 16,260 sessions x 2.302 attempts x 4.124 sec / 60 x $0.016 = $41.16 | $44.99 | $41.16 | -$3.83-8.517% | |
| Guided lesson prompt audio storage and deliveryAWS S3 + CDN audio CloudFront -> guided-lessons assets -> local cache. Live defaults: 878,416,628 stored bytes and 1,820,705,213 bytes downloaded month-to-date. | S3 Standard ap-southeast-2 storage plus CloudFront egress for dqztncf0oxx31.cloudfront.net/guided-lessons.AWS S3 + CloudFront | 0.818 GB x $0.025 + max(0, 1.696 GB - 1,024 GB) x $0.12 = $0.02 | Loading actual usage Stored assets0.778GB Storage / GB0.025$ Month-to-date CDN egress1.849GB CloudFront free egress1,024GB APAC egress / GB0.12$ 0.778 GB x $0.025 + max(0, 1.849 GB - 1,024 GB) x $0.12 = $0.02 | $0.02 | $0.02 | -$0.00-4.936% |
Guided Lesson - Dialogue
Dialogue line -> learner speaks -> pronunciation score -> result shown
Actual vs planned$57.45+$1.11
| Costing sub-feature | Pricing and source | Planned usage | Actual usage | Planned cost | Actual cost | Delta | Note |
|---|---|---|---|---|---|---|---|
| Dialogue sentence pronunciation scoringGoogle Cloud Speech-to-Text V2, chirp_3 Learner sentence -> speech scoring -> result. Cost = sessions x attempts x seconds. | $0.016 / processed minute, first 500k minutesGoogle STT | 14,200 sessions x 3.1 attempts x 4.8 sec / 60 x $0.016 = $56.35 | Loading actual usage Monthly sessions14,122sessions Attempts / session3.331attempts Seconds / attempt4.58sec Rate / min0.016$ 14,122 sessions x 3.331 attempts x 4.58 sec / 60 x $0.016 = $57.45 | $56.35 | $57.45 | +$1.11+1.967% |
Drill
Drill prompt -> learner speaks -> pronunciation score; optional word lookup
Actual vs planned$41.64-$3.78
| Costing sub-feature | Pricing and source | Planned usage | Actual usage | Planned cost | Actual cost | Delta | Note |
|---|---|---|---|---|---|---|---|
| Speaking drill pronunciation scoringGoogle Cloud Speech-to-Text V2, chirp_3 Learner audio -> speech scoring -> result. Cost = sessions x attempts x seconds. | $0.016 / processed minute, first 500k minutesGoogle STT | 12,400 sessions x 3.9 attempts x 3.4 sec / 60 x $0.016 = $43.85 | Loading actual usage Monthly sessions14,697sessions Attempts / session3.452attempts Seconds / attempt2.973sec Rate / min0.016$ 14,697 sessions x 3.452 attempts x 2.973 sec / 60 x $0.016 = $40.21 | $43.85 | $40.21 | -$3.63-8.282% | |
| Drill dictionary lookupCerebras gpt-oss-120b Word tap -> AI explanation -> learner reads. Cost = lookup taps x tokens. | $0.35/M input tokens + $0.75/M output tokensCerebras | 3,200 calls x (850 input / 1M x $0.35 + 260 output / 1M x $0.75) = $1.58 | Loading actual usage Monthly lookups2,965calls Input / lookup891tokens Output / lookup224tokens Input / 1M0.35$ Output / 1M0.75$ 2,965 calls x (891 input / 1M x $0.35 + 224 output / 1M x $0.75) = $1.42 | $1.58 | $1.42 | -$0.15-9.693% |
Roleplay / Conversation Practice
AI speaks -> learner replies -> AI responds -> optional help, review, pronunciation, translation
Actual vs planned$545+$118
| Costing sub-feature | Pricing and source | Planned usage | Actual usage | Planned cost | Actual cost | Delta | Note |
|---|---|---|---|---|---|---|---|
| Shared assumptionsEdit these once. Cost rows below reference them without repeating the inputs. | |||||||
| Session volumeUsage model Users -> sessions -> API calls. No direct cost. | No vendor charge; drives rows belowModel input | Usage volume Session call count Gemini STT + LLM assumptions Google TTS assumptions 2,900 users x 3.3 sessions/user = 9,570 sessions; x 10 calls/session = 95,700 calls | Loading actual usage Usage volume Monthly active users3,228users Sessions / user3.913sessions Session call count STT+LLM calls / session11.4calls TTS calls / session10.4calls Gemini STT + LLM assumptions Gemini text tokens / call19.1tokens Gemini audio sec / call6.258sec Gemini audio tokens / sec37.5tokens Gemini output tokens / call93.8tokens Google TTS assumptions Google TTS chars / call129chars Google TTS rate / 1M chars30$ 3,228 users x 3.913 sessions/user = 12,631 sessions; x 11.4 calls/session = 143,785 calls | Assumption | Assumption | Input only | |
| Session costsOpen session -> AI generates first reply (LLM) -> AI speaks (TTS) -> learner records -> AI understands and replies (STT+LLM) -> AI speaks again (TTS) -> repeat until goals are done. | |||||||
| STT + LLM conversation callsGemini 2.5 Flash-Lite Session -> 8-12 STT+LLM calls. Includes text, audio, and output tokens. | $0.10/M text input + $0.30/M audio input + $0.40/M outputGemini | 9,570 sessions x 10 calls/session x (20 text tokens / 1M x $0.1 + 6 sec x 32 audio tokens/sec / 1M x $0.3 + 105 output tokens / 1M x $0.4) = $9.72 (95,700 calls) | Loading actual usage 12,631 sessions x 11.4 calls/session x (19.1 text tokens / 1M x $0.1 + 6.258 sec x 37.5 audio tokens/sec / 1M x $0.3 + 93.8 output tokens / 1M x $0.4) = $15.78 (143,785 calls) | $9.72 | $15.78 | +$6.06+62.3% | |
| Assistant speech outputGoogle Cloud Text-to-Speech, Chirp 3 HD Session -> AI replies -> speech audio. | $30 / 1M synthesized charactersGoogle TTS | 9,570 sessions x 10 TTS calls/session x 140 chars/call / 1M x $30 = $402 | Loading actual usage 12,631 sessions x 10.4 TTS calls/session x 129 chars/call / 1M x $30 = $507 | $402 | $507 | +$105+26.2% | |
| Goal completion checksGemini 2.5 Flash-Lite One check per goal. Default assumes 4 goals/session. | $0.10/M text input + $0.40/M outputGemini | 2,900 users x 3.3 sessions/user x 4 events/session x (650 input / 1M x $0.1 + 80 output / 1M x $0.4) = $3.71 | Loading actual usage Checks / session3.943checks Input / check557tokens Output / check81.8tokens 3,228 users x 3.913 sessions/user x 3.943 events/session x (557 input / 1M x $0.1 + 81.8 output / 1M x $0.4) = $4.40 | $3.71 | $4.40 | +$0.69+18.6% | |
| Optional action costsTriggered only when the learner asks for extra help. | |||||||
| Hint generationGemini 2.5 Flash-Lite Hint tap -> short suggestion. | $0.10/M text input + $0.40/M outputGemini | 2,900 users x 3.3 sessions/user x 0.75 events/session x (500 input / 1M x $0.1 + 120 output / 1M x $0.4) = $0.70 | Loading actual usage Hints / session0.721hints Input / hint420tokens Output / hint106tokens 3,228 users x 3.913 sessions/user x 0.721 events/session x (420 input / 1M x $0.1 + 106 output / 1M x $0.4) = $0.77 | $0.70 | $0.77 | +$0.07+9.303% | |
| Azure pronunciation checkAzure Pronunciation Assessment Pronunciation tab -> audio check -> score. | Modeled as $1.00 / audio hourAzure Speech | 2,900 users x 3.3 sessions/user x 0.2 checks/session x 8 sec / 3600 x $1 = $4.25 | Loading actual usage Checks / session0.226checks Audio sec / check9.684sec Rate / hour1$ 3,228 users x 3.913 sessions/user x 0.226 checks/session x 9.684 sec / 3600 x $1 = $7.69 | $4.25 | $7.69 | +$3.43+80.7% | |
| TranslationGroq, openai/gpt-oss-120b Translate tap -> English support text. | $0.15/M input + $0.60/M outputGroq | 2,900 users x 3.3 sessions/user x 1.2 events/session x (650 input / 1M x $0.15 + 180 output / 1M x $0.6) = $2.36 | Loading actual usage Translations / session1.147translations Input / translation555tokens Output / translation169tokens 3,228 users x 3.913 sessions/user x 1.147 events/session x (555 input / 1M x $0.15 + 169 output / 1M x $0.6) = $2.68 | $2.36 | $2.68 | +$0.32+13.5% | |
| Final feedback generationGemini / Vertex fallback Session ends -> AI summarizes performance. | $0.10/M text input + $0.40/M outputGemini | 2,900 users x 3.3 sessions/user x 1 events/session x (2,200 input / 1M x $0.1 + 700 output / 1M x $0.4) = $4.79 | Loading actual usage Feedback / session0.902events Input / feedback2,674tokens Output / feedback768tokens 3,228 users x 3.913 sessions/user x 0.902 events/session x (2,674 input / 1M x $0.1 + 768 output / 1M x $0.4) = $6.55 | $4.79 | $6.55 | +$1.76+36.9% | |
Flashcard Drill
Saved card -> learner speaks -> pronunciation score -> result shown
Actual vs planned$48.08+$11.05
| Costing sub-feature | Pricing and source | Planned usage | Actual usage | Planned cost | Actual cost | Delta | Note |
|---|---|---|---|---|---|---|---|
| Flashcard pronunciation scoringGoogle Cloud Speech-to-Text V2, chirp_3 Card prompt -> learner audio -> score. Cost = sessions x cards x seconds. | $0.016 / processed minute, first 500k minutesGoogle STT | 6,200 sessions x 8 attempts x 2.8 sec / 60 x $0.016 = $37.03 | Loading actual usage Monthly drill sessions6,973sessions Cards / session9.638cards Seconds / card2.683sec Rate / min0.016$ 6,973 sessions x 9.638 attempts x 2.683 sec / 60 x $0.016 = $48.08 | $37.03 | $48.08 | +$11.05+29.8% |