One line
opentoken.init()
Auto-instruments your framework and fails open: if we're unreachable, calls go straight to Bedrock.
Plug in. See every agent.
One line, OpenTelemetry, or no code at all for Bedrock.
Your frameworks, out of the box
OpenAI SDK
1 line
LangChain
1 line
LangGraph
OpenTelemetry
LlamaIndex
1 line
Strands Agents
OpenTelemetry
Bedrock AgentCore
No code
boto3
SDK
Voice agents, per minute
LiveKit calls: cost per call and per minute, and latency for each stage.
US$0.031 / min
Works with any OpenTelemetry
If it emits OpenTelemetry, it shows up with cost.
Four ways to connect
Change one line
Costs, alerts and traces per call.
OpenTelemetry
Full agent traces, joined to cost.
One-line SDK
Auto-instrumented and fail-open.
No code
Import from your AWS account.
Deployment modes
SaaS gateway
Real-time alerts; content not stored unless you opt in.
Out of band
Nothing in your request path: zero impact on critical channels.
Data plane in your account
On Enterprise: only metadata aggregates leave your AWS account.
Step-by-step guides are in the customer docs.Docs (customers)
Send data where your teams look
Langfuse and LangSmith
Traces for evals.
Datadog and CloudWatchSoon
Your observability.
Your S3
Bulk exports to your bucket.
FOCUS cost exports
The FinOps standard, on Business and Enterprise.
Items marked Soon are on our public roadmap and not yet generally available.
See your AI spend the way finance does
Accounts are set up by our team. Tell us which model providers you use and we'll walk you through a live workspace with your own numbers.