A
AgentCosts.xyz

AgentCosts Router

Know where your AI budget goes before the invoice hurts.

AgentCosts Router is a cost-first AI gateway for indie builders. It tracks workflow spend, sends weekly savings reports, and warns you before budgets drift.

Weekly waste found

$83
Budget used
74%
ExtractionDeepSeek
Realtime UXGroq
Hard reasoningOpenAI
$50-$500
target monthly spend
$9/mo
indie builder plan
Weekly
savings report

Provider dashboards show spend after it happens, not before a budget breaks.

Indie builders rarely know which workflow is burning the expensive model.

Routine extraction, tagging, and classification calls need cost reports before routing automation.

MVP Loop

One gateway, one report, one budget alert at a time.

01

Connect one gateway

Use one beta ingest key to log provider, model, tokens, workflow, and estimated cost.

02

Get a savings report

See which workflows need premium reasoning and which ones are likely wasting budget.

03

Set budget alerts

Get warned before the invoice hurts, then graduate into cheap, fast, and premium routing rules.

Token Cost Model

Convert usage into spend before requesting onboarding.

Request early access

Token Cost Calculator

Convert tokens into real model spend.

Estimate LLM cost from input, cached input, output tokens, and request volume. Switch to budget mode to see how far a fixed spend can go.

Definition: Lead scoring, intent labels, moderation flags, routing decisions, and other short-label tasks.

Token preset: Small prompt, very short output, high request volume.

Workload profiles only prefill token/request assumptions. They do not decide model quality or peer comparison groups.

Estimated total cost

$6.048

Cost per request

$0.00012096

Weighted cost / 1M tokens

$0.1234

Cost / 1K tokens

$0.00012343

Billable breakdown

Uncached input35M tokens$4.90
Cached input10M tokens$0.0280
Output4M tokens$1.12

Current model

deepseek-v4-flash

$6.048

Comparable route

glm-4.7-flashx

$4.15

Comparison basis

Routine low-cost peer

Filtered by model capability tier first, then by price for this token shape. This is not a global cheapest-model ranking.

Potential savings

$1.898

Pricing source: DeepSeek · Reviewed: 2026-05-21

Legacy deepseek-chat and deepseek-reasoner map to DeepSeek V4 Flash modes.

Prices are public list prices for text tokens and can change. Verify provider billing pages before production billing decisions.

Track this spend in Router

Product Direction

The directory stays useful. The product is Router.

Public pricing data remains the acquisition layer. Router turns that insight into a workflow: track real calls, report waste, alert on budget risk, and only then enable simple routing lanes.

Request Router onboarding