Know where every
AI dollar goes
SuperPenguin tracks cost per request, per customer, per prompt version across OpenAI and Anthropic. One line of code. Five-minute setup.
Works with the providers you already use
What you get
Cost visibility down to the request
Both providers. One screen.
OpenAI and Anthropic aggregated into a single dashboard with automatic sync. Spend by model, by project, by day. Month-over-month trends and a rolling 14-day forecast.
Total Spend
$12,847
+14% vs last month
Total Tokens
In
4.2M
Out
1.8M
Month Forecast
$18,240
based on 14-day avg
By Provider
Daily Spend
Tag every request. Slice any dimension.
Every request is automatically logged with cost, latency, and any metadata you attach. Filter and drill down by any combination in the attribution dashboard.
acme_corp
support_agent
gpt-4o
ticket_summary v3
globex_inc
doc_search
claude-sonnet
search_query v1
acme_corp
summarizer
gpt-4o-mini
brief_summary v2
Compare prompt versions by cost.
Track spend per prompt key across versions. See which iteration is cheaper, faster, or both.
ticket_summary
search_query
Proxy costs vs. actual invoices.
Compare what the proxy estimated against what each provider actually billed. Catch discrepancies by provider, model, or day.
Proxy Spend
$12,847
Billed Spend
$12,903
Drift
-0.4%
Within tolerance. No action needed.
Setup
One line of code. Five minutes.
Point your SDK at SuperPenguin. Tag your requests. Open the dashboard.
# Step 1: Swap the base URL
client = OpenAI(
base_url="https://proxy.carrotlabs.ai/v1"
)
# Step 2: Tag any request with metadata
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": prompt}],
extra_body={
"metadata": {
"customer_id": "acme_corp",
"feature": "support_agent",
"team": "cx",
"environment": "production",
"prompt_key": "ticket_summary",
"prompt_version": "3"
}
}
)Two paths in
Start with billing sync. Add the proxy when ready.
Spend Dashboard
Plug in your provider admin key. We sync billing data automatically. Total spend, model breakdown, daily trends, invoice reconciliation.
- Both providers in a single view
- Automatic billing sync
- 14-day rolling forecast
- Invoice reconciliation
Attribution Proxy
Swap your base URL. Requests flow through SuperPenguin, get tagged with cost telemetry and your metadata, then forward unchanged.
- Per-request cost in real time
- Tag by customer, feature, team, prompt
- Works with the stock OpenAI SDK
- Supports OpenAI and Anthropic
Security
Zero lock-in. Zero data exposure.
No prompts stored. Ever.
The proxy logs token counts, model, latency, and your metadata. Prompt content and responses never touch our servers.
Sub-10ms overhead
Async telemetry. Less than 10ms on non-streaming, less than 5ms TTFT impact on streaming. Your users won't notice.
Your keys pass through
Provider credentials stay in the Authorization header. We never store, log, or touch your API keys.
Pricing
Simple, usage-based pricing.
Based on the AI spend you manage through SuperPenguin.
Free
$0 / mo
For individuals getting started
- Up to $2K managed spend
- All dashboard features
- Attribution proxy
- Reconciliation
- Email support
Growth
$30 / mo
For small teams shipping AI
- Up to $5K managed spend
- Everything in Free
- 3 team members
- Slack cost alerts
- Priority support
Pro
Most popular$200 / mo
For teams scaling AI in production
- Up to $20K managed spend
- Everything in Growth
- 10 team members
Enterprise
Custom
For organizations with high volume
- $20K+ managed spend
- Everything in Pro
- Unlimited team members
- SAML SSOComing soon
Questions
Your AI bill shouldn't be
a mystery.
Per-request, per-customer, per-prompt cost visibility in five minutes.
Start tracking for freeNo credit card. No meetings. No vendor lock-in.