Costs reconciled with your AWS invoice (CUR)
Real cost with your discounts, not list prices.
Estimated at list prices100%
Invoice · AWS Cost and Usage Report88%
−12%
Your discount applied
Chargeback and FOCUS exports on Business and Enterprise
AI CONSUMPTION MONITORING & GOVERNANCE — ACROSS 9 MODEL PROVIDERS
Open Token — Every AI dollar, accounted for and under control.
Spend by team, feature and customer, with agent dashboards, traces and runaway alerts. Bring your own keys for 9 model providers, Amazon Bedrock included.
BYOKYour provider bills your models
PLANOpen Token: subscription + usage
MARKUPNone on tokens
KEYSNo AWS keys required
Acme Bank/Overview
Sep 2026
AI spend this month
US$48,210
Reconciled with your AWS invoice
Daily spendAnomaly
$0$1k$2k$3kSep 1Sep 8Sep 15Sep 22Sep 30SEP 19 · LOOP
Trace · support-triage run
US$0.023
- Classify ticketUS$0.004
- Search knowledge baseUS$0.001
- Draft replyUS$0.015
- Policy checkUS$0.003
Draft reply is 65% of this run’s cost
Loop detected in support-triageCustomer support team alerted · trace linkedOpen trace
9
model providers
Bring your own key. Amazon Bedrock included.
100%
of calls metered and attributed
By team, feature and customer.
10
seats included
No per-seat pricing.
BRING YOUR OWN KEY
Your keys. Your provider deals.
Open Token is BYOK only, across 9 providers: Amazon Bedrock, OpenAI, Anthropic, Google Gemini, Vertex AI, Azure OpenAI, Mistral, Cohere and any OpenAI-compatible endpoint. Each provider bills you directly, so everything you negotiated still applies.
Your apps and agents
One line, OpenTelemetry, or no code at all for Bedrock.
Open Token
MeteringAttributionAlerts
Meters and attributes usage. Never sells it.
Open Token console
Metadata only: cost, tokens, latency, attribution
YOUR AWS ACCOUNT
INFERENCE NEVER LEAVES
Amazon BedrockYour pricing, AWS credits and commitmentsINFERENCE
Audit trailEvery access by us, loggedYOURS
Cost reportsWith our attribution tagsCUR
- Request path, through a role you can revoke
- Metadata to the console
- Read-only cost reports
Inference runs on your accounts
Requests use the connection you choose: Amazon Bedrock in your AWS account, or your key at another provider. Never ours.
Keep your pricing
Your discounts and commitments with each provider, and your AWS credits, keep working. No markup.
We never touch your tokens
We don't sell model usage or hold a balance. We meter and attribute it.
No AWS keys required
On Bedrock, a narrowly scoped role you create, revocable anytime. Other providers' keys are stored encrypted, never shown again.
PRICING
What you pay
Two bills, never mixed. Your providers bill your models at your prices. Open Token bills a subscription plus usage.
BILLED BY YOUR PROVIDERS
Your models
BYOK, at your own provider prices
Model usageYour rates
AWS credits & commitmentsApplied
Token markupUS$0.00
BILLED BY OPEN TOKEN
The platform
All included in the platform:
BYOK gateway
Agent observability
Spend control
Finance
+US$3.00per 1,000 traces
Subscription plus usage. Only stored AI traces count.
AGENT OBSERVABILITY
Everything finance and engineering need, on one screen.
Built around your agents, not around log lines.
Acme Bank/Agents/support-triage
ILLUSTRATIVE DATA · SEP 2026
A dashboard for every agent
support-triage · runs per day18,240 runs
Traces with cost on every step
4 steps · 2.4 sUS$0.023
00.6 s1.2 s1.8 s2.4 s
- MODELClassify ticketUS$0.004
- TOOLSearch knowledge baseUS$0.001
- MODELDraft replyUS$0.015
- MODELPolicy checkUS$0.003
Draft reply made this run expensive: 65% of its cost.
Runaway agents, caught mid-run
LOOP · SEP 19, 14:02
- ×1search_knowledge_base(args: same)
- ×2search_knowledge_base(args: same)
- ×3search_knowledge_base(args: same)
Customer support alerted, with a link to the trace
Automatic stop in the request pathSOON
Cost, runs and cost per run for each project and agent.
See exactly which call made a run expensive.
Loops and threshold breaches alert the owning team with a link to the trace.
FINANCE & OPERATIONS
Real cost, reconciled with your AWS invoice.
Your discounts applied, not list prices. Then the guardrails that keep critical channels fast and auditors satisfied.
Proof for auditors
Every access by Open Token lands in your own AWS audit trail.
your-aws-account / audit trail
- ✓Role assumed by Open Token09:14:02
- ✓Model invoked in us-east-109:14:03
- ✓Cost report read09:30:00
Voice agents
LiveKit calls with cost per call and per minute, and latency for each stage.
US$0.031/ min
- Speech to text180 ms
- Model420 ms
- Text to speech150 ms
Zero impact on critical channels
The SDK fails open and out-of-band modes stay off the request path.
SDKYOUR APPmetadata, asyncOpen TokenOUT-OF-BANDREQUEST PATH · UNCHANGEDBedrockYOUR AWSFails open if unreachable
WHY OPEN TOKEN
Keep your eval tools. Add the spend layer.
Open Token adds AWS-reconciled cost, attribution and budget alerts next to the tools your engineers already use.
KEEP
Evals, prompts and tracing
Langfuse, LangSmith, or whatever your team already trusts
ADD
Open Token — the spend layer
AWS-reconciled costAttributionBudgets & alertsAgent traces
YOURS
Your model providers, with your keys
BILLED BY YOUR PROVIDERS
Your models, your negotiated pricing
Open Token
Spend layer across 9 model providers: AWS-reconciled Bedrock cost, attribution, budgets and runaway alerts
Subscription + US$3.00 per 1,000 traces · 10 seats included
LangSmithTracing, evals, datasets and prompts, plus an LLM gateway with spend policies
From US$39 per seat / month
LangfuseOpen-source tracing, evals and prompt management, cloud or self-hosted
Cloud from US$29 / month, unlimited users
PortkeyMulti-provider AI gateway with budgeted virtual keys
Production from US$49 / month
HeliconeProxy-based request logging, caching and rate limits
Maintenance mode since March 2026
Public documentation, September 2026.
Detailed comparisonsWORKS WITH YOUR STACK
One line, OpenTelemetry, or no code at all for Bedrock.
Python
Node.js
OpenAI SDK
boto3
LangChain
LangGraph
LlamaIndex
Vercel AI SDK
Strands Agents
Bedrock AgentCore
LiveKit Agents
OpenTelemetrySee your AI spend the way finance does.
Tell us which model providers you use and we'll walk you through a live workspace with your own numbers.
Tell us about your setup
Which model providers your teams and agents use today.
Our team sets up your account
Accounts are set up by our team, with a role you control.
A live workspace, your numbers
Spend by team, feature and customer.