Torque / Model manager & usage

Model manager & usage

Which AI models the assistant runs on, what each one costs, and where the month's budget is going. Pick the default route, watch per-model token spend, and keep the workshop's running cost in check. Live, light + dark, built only from registered primitives.

Production answer

Model manager & usage is a reusable Oak Flats Muffler Men UI primitive with documented states, accessibility expectations, theme behavior, and implementation evidence.

Primary CTAReview Model manager & usage states
Generative search brief

Model manager & usage: Which AI models the assistant runs on, what each one costs, and where the month's budget is going. Pick the default route, watch per-model token spend, and keep the workshop's running cost in check. Live, light + dark, built only from registered primitives.

Torque · Model managerOak Flats Muffler MenIllawarra · NSW South Coast
Default model

New jobs route to Torque Shop · 200K ctx · $3.00 / 1M tokens

Reporting period

This reporting period — 1–28 May 2026

Usage at a glance

What Torque is spending

Tokens against the monthly budget, the dollar cost, how many jobs were routed, and the share the assistant handled without the crew — measured over the reporting period.

Live · synced
Available models

Every model the assistant can route to

Four tiers, one budget. The workshop runs day-to-day on the Sonnet-class default and only escalates to the deep-reasoning model for quotes and tricky diagnostics — that is what keeps the cost per handled lead at thirteen cents.

  • Torque ShopDefault

    Front-of-house · default route

    Live
    Tokens4,120,000 / 6,000,000
    Requests
    5,840
    Cost (28d)
    $24.10
    Per 1M
    $3.00
    7-day requestsTorque Shop request volume over the last 7 daysTrend over 7 samples ranging from 120.0 to 181.0.

  • Torque Quick

    After-hours auto-reply · SMS

    Live
    Tokens1,980,000 / 3,000,000
    Requests
    9,310
    Cost (28d)
    $4.30
    Per 1M
    $0.80
    7-day requestsTorque Quick request volume over the last 7 daysTrend over 7 samples ranging from 40.0 to 78.0.

  • Torque Draft

    Content drafting · long context

    Live
    Tokens1,410,000 / 2,400,000
    Requests
    412
    Cost (28d)
    $8.40
    Per 1M
    $0.35
    7-day requestsTorque Draft request volume over the last 7 daysTrend over 7 samples ranging from 18.0 to 41.0.

  • Torque Reason

    Quotes & diagnostics · escalation only

    Fallback
    Tokens470,000 / 1,200,000
    Requests
    96
    Cost (28d)
    $5.00
    Per 1M
    $15.00
    7-day requestsTorque Reason request volume over the last 7 daysTrend over 7 samples ranging from 6.0 to 14.0.
Token & cost trends

Where the tokens and dollars go

Token consumption split by model over twelve weeks, the dollar cost per model across the last six reporting weeks, the 28-day budget, and a breakdown of the spend.

Live · synced
Token usage by model12 weeks · millions
Weekly token consumption over twelve weeks, split across the four Torque modelsStacked area chart over 12 steps. 4 series.0M1M2M2M3MW1W2W3W4W5W6W7W8W9W10W11W12

Torque Shop carries most of the load. Torque Draft climbs whenever the content engine writes a batch of service pages — long context, but cheap per token.

Cost by model6 weeks · AUD
Weekly dollar cost per Torque model across the last six reporting weeksGrouped bar chart with 3 series across 6 categories.02356W23W24W25W26W27W28

Reason is the priciest per token, but escalation-only routing keeps its weekly cost under the cheap-and-fast Shop model.

Budget meter28-day budget
Budget health
Token budget 66% of 100Radial meter at 66 percent.66%Token budget
Spend cap 70% of 100Radial meter at 70 percent.70%Spend cap
Cache hit 41% of 100Radial meter at 41 percent.41%Cache hit
14-day spendTorque daily spend over the last fourteen daysTrend over 14 samples ranging from 1.2 to 2.0.$41.80 of $60 · 66% of the token budget used
Spend breakdownBy model
Drilldown · Model

Spend this period

Total $41.80 across 1–28 May 2026, 15,658 requests routed

Torque Shop · front-of-house
$24.1058%
Torque Draft · content
$8.4020%
Torque Reason · quotes
$5.0012%
Torque Quick · after-hours
$4.3010%
4 contributorsTop Torque Shop · front-of-house
Period over periodEfficiency
This period$0.13
24% cheaper
Prior period$0.17
Cost per handled lead trendTrend over 7 samples ranging from 0.1 to 0.2.
This period41%
+12pt
Prior period29%
Prompt-cache hit rate trendTrend over 7 samples ranging from 27.0 to 41.0.
Torque Shop0.8s
38% faster
Prior period1.3s
Median first-token trendTrend over 7 samples ranging from 0.8 to 1.3.
Per-model detail

Usage by model

This reporting period
Sorted by tokensToken, request and cost usage per Torque model
Token, request and cost usage per Torque model
ModelTierStatus7d tokens
Torque ShopDefault · front-of-houseSonnet4.12M5,840$24.10LiveTorque Shop token usage over the last 7 daysTrend over 7 samples ranging from 120.0 to 181.0.
Torque QuickAfter-hours · SMSHaiku1.98M9,310$4.30LiveTorque Quick token usage over the last 7 daysTrend over 7 samples ranging from 40.0 to 78.0.
Torque DraftContent · long contextFlash1.41M412$8.40LiveTorque Draft token usage over the last 7 daysTrend over 7 samples ranging from 18.0 to 41.0.
Torque ReasonQuotes · escalation onlyOpus0.47M96$5.00FallbackTorque Reason token usage over the last 7 daysTrend over 7 samples ranging from 6.0 to 14.0.