Cost & Usage Dashboard

The Cost & Usage dashboard gives super-admins a unified view of AI and Twilio spending across the platform. It aggregates AI usage logs, Twilio voice/SMS/phone number costs, and renders breakdowns by model, provider, feature, and organization — with sparkline trends, efficiency metrics, and BYO Twilio distinction.

API Endpoint

GET /api/admin/ai-costs?time_range=<TimeRange>

Auth: Super-admin only (checked via ensure_super_admin).

Query parameter: time_range — a TimeRange enum value (Last7Days, Last30Days, Last90Days, AllTime).

Response: AdminAiCostsData containing KPIs, breakdowns, time-series data, and Twilio costs.

Response Shape

pub struct AdminAiCostsData {
    // AI cost KPIs
    pub total_cost: f64,
    pub previous_total_cost: f64,
    pub cost_change_pct: f64,
    pub total_api_calls: u64,
    pub avg_cost_per_call: f64,
    pub cost_per_day_avg: f64,
    pub top_model: String,
    pub top_model_cost: f64,

    // Token totals
    pub total_input_tokens: u64,
    pub total_output_tokens: u64,

    // Breakdowns
    pub cost_by_model: Vec<AdminAiCostBreakdownItem>,
    pub cost_by_org: Vec<AdminAiCostOrgItem>,
    pub cost_by_provider: Vec<AdminAiCostBreakdownItem>,
    pub cost_by_feature: Vec<AdminAiCostBreakdownItem>,

    // Time series — single line
    pub cost_series: Vec<AdminCostChartPoint>,

    // Grouped time series — stacked charts
    pub cost_series_labels: Vec<String>,
    pub cost_series_by_model: Vec<AdminCostSeriesGroup>,
    pub cost_series_by_feature: Vec<AdminCostSeriesGroup>,
    pub cost_series_by_provider: Vec<AdminCostSeriesGroup>,

    // Twilio
    pub twilio: AdminTwilioCostsData,

    // Combined total
    pub grand_total_cost: f64,

    pub time_range: TimeRange,
    pub range_label: String,
}

Breakdown Items

Each breakdown item carries cost, percentage, and optional token/efficiency metrics:

pub struct AdminAiCostBreakdownItem {
    pub label: String,
    pub cost: f64,
    pub percentage: f64,
    pub api_calls: Option<u64>,
    pub input_tokens: Option<u64>,
    pub output_tokens: Option<u64>,
    pub sub_items: Option<Vec<AdminAiCostBreakdownItem>>,
    pub trend: Option<Vec<f64>>,
    pub cost_per_1k_tokens: Option<f64>,
    pub cache_hit_pct: Option<f64>,
}

• trend — sparkline values aligned with time-series buckets
• cost_per_1k_tokens — efficiency metric: total cost ÷ total tokens × 1000
• cache_hit_pct — cached tokens ÷ (input + cached tokens)
• sub_items — nested breakdown (e.g., models within a feature)

Per-Organization Items

pub struct AdminAiCostOrgItem {
    pub organization_id: Uuid,
    pub organization_name: String,
    pub total_cost: f64,
    pub openrouter_cost: f64,
    pub openai_cost: f64,
    pub gemini_cost: f64,
    pub top_feature: String,
    pub api_calls: u64,
    pub input_tokens: u64,
    pub output_tokens: u64,
}

Grouped Time Series

pub struct AdminCostSeriesGroup {
    pub label: String,       // e.g., "GPT-4o", "Voice Agent", "OpenAI"
    pub values: Vec<f64>,    // Aligned with cost_series_labels
}

The backend selects the top 5 groups by total cost and aggregates the rest into an "Other" group.

Twilio Costs

The dashboard aggregates Twilio expenses from call, message, phone_number, and organization_twilio_settings tables.

Response Shape

pub struct AdminTwilioCostsData {
    pub voice_total_minutes: f64,
    pub voice_total_cost: f64,
    pub voice_cost_per_minute: f64,
    pub voice_call_count: u64,
    pub phone_number_count: u64,
    pub phone_number_monthly_cost: f64,
    pub sms_count: u64,
    pub sms_total_cost: f64,
    pub sms_inbound_count: u64,
    pub sms_outbound_count: u64,
    pub total_twilio_cost: f64,
    pub loquent_twilio_cost: f64,
    pub cost_by_org: Vec<AdminTwilioCostOrgItem>,
}

Per-Organization Twilio

pub struct AdminTwilioCostOrgItem {
    pub organization_id: Uuid,
    pub organization_name: String,
    pub voice_minutes: f64,
    pub voice_cost: f64,
    pub call_count: u64,
    pub phone_numbers: u64,
    pub phone_number_cost: f64,
    pub sms_count: u64,
    pub sms_cost: f64,
    pub is_byo: bool,
}

BYO vs Loquent-Managed

Organizations using their own Twilio credentials (is_byo: true in organization_twilio_settings) are tracked separately:

• total_twilio_cost — all organizations, including BYO
• loquent_twilio_cost — excludes BYO organizations
• grand_total_cost — AI cost + loquent_twilio_cost (reflects actual platform spend)

BYO organizations display a badge in the Twilio org table.

Twilio Pricing Constants

Defined in src/mods/admin/services/twilio_pricing.rs:

Constant	Value	Description
TWILIO_VOICE_PER_MINUTE	$0.014	US programmable voice
TWILIO_PHONE_NUMBER_MONTHLY	$1.15	Local US number monthly fee
TWILIO_SMS_OUTBOUND_PER_SEGMENT	$0.0079	Outbound SMS per segment
TWILIO_SMS_INBOUND_PER_SEGMENT	$0.0075	Inbound SMS per segment

Phone number cost is prorated by the selected time range: count × $1.15 × (days_in_range / 30).

AI Pricing Engine

The get_model_pricing() function maps (provider, model) pairs to per-1M-token rates. The calculate_cost() function applies these rates to a usage log row’s token counts.

pub struct ModelPricing {
    pub input_per_1m: f64,
    pub output_per_1m: f64,
    pub cached_per_1m: f64,
    pub audio_input_per_1m: f64,
    pub audio_output_per_1m: f64,
    pub audio_per_minute: f64,
}

Supported models and rates:

Provider	Model	Input	Output	Audio In	Audio Out
any	deepseek/deepseek-v3.2	$0.25	$0.40	—	—
any	google/gemini-3.1-pro-preview	$2.00	$12.00	—	—
any	google/gemini-3.1-flash-lite-preview	$0.25	$1.50	—	—
any	anthropic/claude-sonnet-4.6	$3.00	$15.00	—	—
openai	gpt-realtime-mini	$0.60	$2.40	$10.00	$20.00
openai	gpt-realtime / gpt-realtime-1.5	$4.00	$16.00	$32.00	$64.00
gemini	gemini-live-2.5-flash-native-audio	$0.50	$12.00	—	—
openai	gpt-4o-transcribe	$2.50	$10.00	$6.00	—

Unknown models fall back to zero-cost with a warning log.

Cost formula:

cost = (input_tokens × input_rate
      + output_tokens × output_rate
      + cached_tokens × cached_rate
      + input_audio_tokens × audio_input_rate
      + output_audio_tokens × audio_output_rate) / 1,000,000
      + audio_duration_secs / 60 × audio_per_minute

Dashboard Layout

The UI renders six sections:

1. KPI cards (3×2 grid):

   • Grand Total (AI + Twilio)
   • AI Spend (with period-over-period change)
   • Twilio Spend (Loquent-managed only if BYO orgs exist)
   • API Calls
   • Tokens Used (input/output breakdown)
   • Voice Minutes (call count and cost)

2. Grouped cost chart — filterable stacked bar chart with modes: Total, By Model, By Feature, By Provider

3. Breakdown cards (3 columns) — Model, Provider, and Feature breakdowns with sparkline trends, token counts, efficiency metrics ($/1K tokens, cache hit %), and nested model details for features

4. AI cost by organization table — per-org totals with API calls, input/output tokens, and provider splits

5. Twilio expense cards (3 cards) — Voice (minutes, cost), Phone Numbers (count, monthly cost), SMS (inbound/outbound counts, cost)

6. Twilio cost by organization table — per-org voice, calls, numbers, SMS, with BYO badge

Adding a New Model

1. Add a match arm in get_model_pricing() at src/mods/ai/types/ai_pricing_type.rs
2. Set the per-1M-token rates from the provider’s pricing page
3. The cost dashboard picks up the new model automatically on next query

Key Files

File	Purpose
src/mods/ai/types/ai_pricing_type.rs	ModelPricing, get_model_pricing(), calculate_cost()
src/mods/admin/services/admin_ai_cost_service.rs	Aggregation queries and breakdown computation
src/mods/admin/services/twilio_pricing.rs	Twilio pricing constants
src/mods/admin/api/get_admin_ai_costs_api.rs	GET /api/admin/ai-costs endpoint
src/mods/admin/types/admin_data_types.rs	Response types
src/mods/admin/components/admin_ai_costs_tab_component.rs	Dashboard UI components