Control AI spend before it happens

Set budgets for the organization, teams, routers, people and models. Get alerts before a limit, block at it, and know what every customer and feature costs.

Budgets for the organization, routers and tags

A budget covers the whole organization, a list of up to 200 routers, or every router carrying one of up to 50 tags. Each budget sets a total cap, a cap on each router and caps on single models, with its own Notify and Block rules.

All budgets + New budget
Status:AllNearingImminentCapped Search budgets…
Total spend$14,286
Effective limits$32,200
Capped1
Imminent1
Nearing1
Unpriced traffic0.3%
Budgets 1–4 of 425 / pageColumnsUpdated
NameScopePeriodSpend / effective limitUsedStatusUpdated
⌄ Production APIbdg_7Hq2Lm4NwQ Tag · env:prod Monthly $8,913 / $10,000 89% Nearing 10/28/2026
Production APIbdg_7Hq2Lm4NwQTag · env:prodMonthly Detailed usage →
Spent$8,913
Effective limit$10,000
Remaining$1,087
Per-router limit$4,000 base
PeriodOct 1 – Oct 31 resets Nov 1
Pending requests0
Routers over cap0 of 4
Spend per day · Daily · Oct 1 – 31Spend
$500$250$0
Oct 1Oct 8Oct 15Oct 22Oct 29
› Support routersbdg_3Kd9WqTz1M Routers · 2 Monthly $1,500 / $1,500 100% Capped 10/27/2026
› Search teambdg_9Tz4RbXc2P Tag · team:search Weekly $641 / $700 92% Imminent 10/24/2026
› Acmebdg_2Mv8XcLp6R Org Monthly $14,286 / $20,000 71% On track 10/21/2026
Limits are server-resolved effective limits for the current UTC period.

Caps for every team, person and model

Scope a budget to the organization, a team, a tag, a router or a person. Each one takes a Total cap, a cap on every line and caps on single models, across the whole budget or inside one line, with Notify and Block rules. Policies create the same budget for every team or person they match.

Caps for any scope same budget shape, every scope Day 30
Budget Acme Org · every router and every call with no router
Period Monthly resets Nov 1, 00:00 UTC
Rules Notify at 80% of the total · email finance@acme.exampleBlock at 100% of each model cap
Owner Created by an admin
1Total cap Everything in the scope, as one pool.
Acme $14,540.00 / $20,000.00 On track
2Per-router caps Each router against its own line.
support-bot $5,100.00 / $6,000.00 Nearing
search-api $4,300.00 / $5,000.00 Nearing
summarize-prod $2,900.00 / $4,000.00 On track
passthroughcalls with no router $2,240.00 / $3,000.00 On track
3Model caps One model across the whole budget, or inside one line.
anthropic/claude-opus-4-1across the whole org $3,000.00 / $3,000.00 At cap
openai/gpt-5in search-api $1,050.00 / $1,200.00 Nearing
This period
Day 24anthropic/claude-opus-4-1 reached its $3,000.00 cap across the whole org. Other models keep serving.
Budget Search team Team · Search, 3 routers
Period Weekly resets Monday, 00:00 UTC
Rules Notify when the total is projected to reach 100% within 2 daysBlock at 100% of each per-router cap
Owner Managed by policy · Team spend The policy creates this budget for every match.
1Total cap Everything in the scope, as one pool.
Search team $626.00 / $700.00 Nearing
2Per-router caps Every router the team owns.
search-api $330.00 / $350.00 Nearing
search-rerank $200.00 / $200.00 At cap
query-rewrite $96.00 / $150.00 On track
3Model caps One model across the whole budget, or inside one line.
openai/gpt-5across the team $280.00 / $300.00 Nearing
cohere/rerank-v3.5in search-rerank $120.00 / $120.00 At cap
This period
Day 24cohere/rerank-v3.5 reached its $120.00 cap in search-rerank. Other models keep serving.
Day 24search-rerank reached its $200.00 cap. The rest of the budget keeps serving.
Budget Staging Tag rule · env:staging, 4 routers
Period Daily resets 00:00 UTC
Rules Block at 100% of the totalBlock at 100% of each per-router cap
Owner Created by an admin
1Total cap Everything in the scope, as one pool.
Staging $120.00 / $120.00 At cap
2Per-router cap from the tag rule $40.00 for every router tagged env:staging, including routers tagged later.
support-bot-stg $38.00 / $40.00 Nearing
search-api-stg $40.00 / $40.00 At cap
checkout-assist-stg $31.00 / $40.00 On track
summarize-stg $27.00 / $40.00 On track
3Model caps One model across the whole budget, or inside one line.
anthropic/claude-opus-4-1across every staging router $15.00 / $15.00 At cap
This period
Day 20anthropic/claude-opus-4-1 reached its $15.00 cap across every staging router. Other models keep serving.
Day 26search-api-stg reached its $40.00 cap. The rest of the budget keeps serving.
Budget Checkout assistant Routers · checkout-assist
Period Monthly resets Nov 1, 00:00 UTC
Rules Notify at 90% of each model cap · webhook pagerduty-billingBlock at 100% of the router cap
Owner Created by an admin
1Total cap Everything in the scope, as one pool.
Checkout assistant $1,380.00 / $1,500.00 Nearing
2Router cap One router, one line.
checkout-assist $1,380.00 / $1,500.00 Nearing
3Model caps One model across the whole budget, or inside one line.
anthropic/claude-sonnet-4-5in checkout-assist $840.00 / $900.00 Nearing
openai/gpt-5-miniin checkout-assist $250.00 / $250.00 At cap
google/gemini-2.5-flashin checkout-assist $120.00 / $200.00 On track
This period
Day 24openai/gpt-5-mini reached its $250.00 cap in checkout-assist. Other models keep serving.
Budget Dana Okafor Person · dana@acme.example, 3 agents and keys
Period Monthly resets Nov 1, 00:00 UTC
Rules Notify at 75% of the total · email dana@acme.exampleBlock at 100% of each model cap
Owner Managed by policy · Engineering seats The policy creates this budget for every match.
1Total cap Everything in the scope, as one pool.
Dana Okafor $343.00 / $400.00 Nearing
2Per-agent caps Each coding agent and key the person uses.
Claude Codeissued agent $236.00 / $250.00 Nearing
Cursorissued agent $88.00 / $100.00 Nearing
Personal keymrk_…h2Qa $19.00 / $50.00 On track
3Model caps One model across the whole budget, or inside one line.
anthropic/claude-opus-4-1across every agent $150.00 / $150.00 At cap
openai/gpt-5in Cursor $41.00 / $60.00 On track
This period
Day 22anthropic/claude-opus-4-1 reached its $150.00 cap across every agent. Other models keep serving.
Every scope takes the same caps, periods and rules. Spend updates every few seconds, and a Block rule on a line turns its cap into a hard stop for that line only.

Soft and hard limits

Every budget can carry soft and hard limits. A soft limit alerts by email or signed webhook when spend reaches a percentage of a cap, or is projected to. A hard limit blocks: callers get HTTP 429 budget_exhausted with Retry-After until the period resets.

Production API Monthly · total cap $10,000.00 Oct 12Oct 24Oct 29
$4,611.20 spent$8,204.75 spent$10,000.00 spent
Soft · 80% Hard · 100%
Under both limits. Requests are served.Soft limit crossed. Alerts sent, requests still served.Hard limit reached. Requests are refused until Nov 1.
Soft limit · Notify Waiting · at 80%Fired Oct 24, 09:12 UTC
At 80% of the total cap → email maya.ortiz@example.com, post to Ops alerts Projected to reach 100% of the total cap within 30 min → post to Ops alerts
Sent Production API reached 80% of its monthly capTo maya.ortiz@example.com · $8,204.75 of $10,000.00 POST budget.threshold_reachedOps alerts · signed x-multi-router-signature-256 · 200
Requests keep flowing. Each rule fires once per period.
Hard limit · Block Not holdingHolding until Nov 1
At 100% of the total cap → block requests At 100% of the anthropic/claude-opus-4-1 cap → block that model
Caller receives
HTTP/1.1 429 Too Many Requests
Retry-After: 259200

{
  "error": {
    "message": "budget bdg_7Hq2Lm4NwQ is exhausted for this period; it resets at 2026-11-01T00:00:00Z",
    "type": "budget_exhausted",
    "code": "budget_exhausted"
  }
}
Refused before any provider call, so a blocked request costs nothing.

Overrides and requests for more

Raise one limit for the rest of the period without editing the budget. Employees ask for more in the Portal, request approvers approve at the amount they choose, and every override ends when the period resets.

LFLena Fischer · Employee · uses search-api
search-api PortalNearing
Spend this month$1,742.18 of $2,000.00$3,000.00
Request more budget Send request
Request sent $3,500.00 monthly · waiting for the budget's request approvers.
Approved at $3,000.00 Maya Ortiz raised the limit until Nov 1. Then it returns to $2,000.00.
To maya@example.com, dev@example.com Lena Fischer asked for more budget on search-api $2,000.00 → $3,500.00 monthly · Search routers
Search routers bdg_9Tz4RbXc2P Tag · team:search Monthly Budgets · admin
Spent $5,184.20
Effective limit $9,000.00
Remaining $3,815.80
Per-router limit $2,000.00 base
Period Oct 1 – Oct 31 resets Nov 1
Request approversRequests on

Emailed when an employee asks for a higher limit in the Portal.

maya@example.com ×dev@example.com ×
2 of 10
Pending requests1
LF Lena Fischer for search-api · Oct 22, 09:14 UTC
Monthly limit $2,000.00 → $3,500.00
DenyApprove
No pending requests. Approved at $3,000.00 by Maya Ortiz.
Budget Overrides · this period Each override replaces one limit until the period ends.
+ Add Budget Override
TargetBase limitEffective limitChangeSet by
search-apiRouter $2,000.00 $3,000.00 +50% Maya Ortizapproved request · Oct 22
anthropic/claude-opus-4-1Model $600.00 $900.00 +50% Maya Ortizadded directly · Oct 18
Total capTotal $8,000.00 $9,000.00 +13% Dev Pateladded directly · Oct 18
Overrides end with the period. On Nov 1 every limit returns to its base.

Every price, every region, your negotiated rates

One model can carry a dozen prices: by vendor, region and variant, by service tier, for cache writes that live five minutes or an hour, for cache reads, long prompts and web searches. Multi-Router prices each request at the exact rates that served it. Set your own negotiated rates and budgets, caps and alerts use them instead of list prices.

anthropic/claude-sonnet-4-5 per 1M tokens
OfferingInputOutputCache readCache write 5 minCache write 1 hourBatch inputInput over 200k
Anthropic standard · global $3.00 $15.00 $0.30 $3.75 $6.00 $1.50 $6.00
Amazon Bedrock global · us-east-1 Negotiated · 15% off list $3.00 $2.55 $15.00 $12.75 $0.30 $0.255 $3.75 $3.188 $6.00 $5.10 $1.50 $1.275 $6.00 $5.10
Amazon Bedrock geo · eu-central-1 Negotiated · 15% off list $3.30 $2.805 $16.50 $14.025 $0.33 $0.2805 $4.125 $3.506 $6.60 $5.61 $1.65 $1.403 $6.60 $5.61
Amazon Bedrock in_region · eu-central-1 Negotiated · 15% off list $3.45 $2.933 $17.25 $14.663 $0.345 $0.2932 $4.313 $3.666 $6.90 $5.865 n/a n/a $6.90 $5.865
Vertex AI standard · europe-west1 $3.30 $16.50 $0.33 $4.125 $6.60 $1.65 $6.60
Anthropic custom · Acme agreement Negotiated rates $3.00 $2.40 $15.00 $12.00 $0.30 $0.24 $3.75 $3.00 $6.00 $4.80 $1.50 $1.20 $6.00 $4.80
Each request is priced at the rates of the offering, region and tier that served it, after any negotiated discount.
Production API Monthly budget · cap $10,000.00 Oct 1 to Oct 30
Spent at catalog list prices $11,497.42 / $10,000.00 Capped · Block rule holding
Spent at Acme rates $8,912.75 / $10,000.00 Nearing · $1,087.25 to the cap
Spend by router support-bot $5,412.40$4,212.18 search-api $3,598.10$2,876.40 summarize-prod $2,486.92$1,824.17

Set your negotiated rates once. Every request is priced at them, so budgets, caps, alerts, Traffic and Detailed usage count what you actually pay.

Know what each customer and feature costs you

Create a router for each customer and feature through the API, tagged with customer and feature, and send a session id and a correlation id with every call. Spend adds up from each request to the workflow run, the feature and the customer, so you know the AI cost behind every account and what each feature costs to run.

Unit economics Acme · October · UTC
AI cost$7,330.56
Tokens1.51B
Tasks57,202
Cost per task$0.1282
Customers over budget1
CustomerTasksTokensAI costOf budgetPer taskMay to Oct
FeatureCustomersTasksTokensAI costPer taskvs September
Inbox triagefeature:inbox-triage 7 12,090 331.1M $1,536.00 $0.1270 +4%
Reply draftingfeature:reply-drafting 7 17,800 510.0M $2,362.48 $0.1327 +9%
Call summariesfeature:call-summaries 6 12,981 416.2M $2,216.69 $0.1708 +31%
Knowledge searchfeature:knowledge-search 7 9,691 257.4M $1,215.40 $0.1254 −6%
AI cost adds up the stored cost of every request on routers tagged with the customer.
Cobalt Logistics Over budget
AI cost$2,911.40
Tokens502.0M
Tasks15,880
Per task$0.1833
Cost by feature
Inbox triage $407.60
Reply drafting $611.39
Call summaries $1,601.27
Knowledge search $291.14
Costliest step Summarize call on anthropic/claude-sonnet-4-5 is 61% of this customer's cost.
Budget customer:cobalt-logistics · Monthly · $2,500 · notify at 80% Over its cap. This budget only notifies. A Block rule would stop this customer's calls at the cap.
Pinecrest Hotels 82% of budget
AI cost$1,962.55
Tokens478.7M
Tasks18,240
Per task$0.1076
Cost by feature
Inbox triage $510.26
Reply drafting $863.52
Call summaries $235.51
Knowledge search $353.26
Costliest step Draft reply on anthropic/claude-sonnet-4-5 is 47% of this customer's cost.
Budget customer:pinecrest-hotels · Monthly · $2,400 · notify at 80% $437.45 left this month.
Linden Schools 76% of budget
AI cost$1,140.80
Tokens248.0M
Tasks12,930
Per task$0.0882
Cost by feature
Inbox triage $353.65
Reply drafting $330.83
Call summaries $91.26
Knowledge search $365.06
Costliest step Answer with sources on google/gemini-2.5-pro is 38% of this customer's cost.
Budget customer:linden-schools · Monthly · $1,500 · notify at 80% $359.20 left this month.
Cedar Veterinary 63% of budget
AI cost$503.27
Tokens117.0M
Tasks4,410
Per task$0.1141
Cost by feature
Inbox triage $100.65
Reply drafting $226.47
Call summaries $125.82
Knowledge search $50.33
Costliest step Draft reply on anthropic/claude-sonnet-4-5 is 44% of this customer's cost.
Budget customer:cedar-vet · Monthly · $800 · notify at 80% $296.73 left this month.
Juniper Clinics 69% of budget
AI cost$412.30
Tokens93.7M
Tasks3,120
Per task$0.1321
Cost by feature
Inbox triage $90.71
Reply drafting $169.04
Call summaries $111.32
Knowledge search $41.23
Costliest step Draft reply on anthropic/claude-sonnet-4-5 is 39% of this customer's cost.
Budget customer:juniper-clinics · Monthly · $600 · notify at 80% $187.70 left this month.
Maple & Finch Legal 72% of budget
AI cost$286.15
Tokens46.2M
Tasks1,412
Per task$0.2027
Cost by feature
Inbox triage $34.34
Reply drafting $108.74
Call summaries $51.51
Knowledge search $91.57
Costliest step Answer with sources on google/gemini-2.5-pro is 29% of this customer's cost.
Budget customer:maple-finch-legal · Monthly · $400 · notify at 80% $113.85 left this month.
Tidewater Realty 57% of budget
AI cost$114.09
Tokens29.3M
Tasks1,210
Per task$0.0943
Cost by feature
Inbox triage $38.79
Reply drafting $52.48
Knowledge search $22.82
Costliest step Draft reply on anthropic/claude-sonnet-4-5 is 41% of this customer's cost.
Budget customer:tidewater-realty · Monthly · $200 · notify at 80% $85.91 left this month.

Put a budget on your AI traffic

Create a budget, set its caps and rules, and see spend against it on every request.